Asylum and Channel crossings in the UK

The Migration Observatory counts around 193,000 people detected reaching the UK in small boats from 2018 to the end of 2025. Adding the 16,300 who arrived in 2026 up to 9 September takes the total past two hundred thousand. Crossings were small until 2020: 10,608 across 2018–2020 together.

The Home Office's official count peaks at 45,774 in 2022. Since 2021, every year has had between 29,000 and 46,000 arrivals. In 2025 around 41,500 people crossed, 13% more than in 2024 and the second-highest year.

By 9 September 2026, arrivals were 47% below the same point of 2025 and 27% below 2024. On the 1 January–22 October line in the chart, 2026 is the lowest year since 2021. In the year to June 2026 the official count was 33,374, 23% lower than the year before. Frontex, counting people who reached the UK plus people stopped from leaving France, puts Channel attempts in January–September 2026 down 43%.

Several candidate causes are documented, and none has been measured on its own:

• Good crossing weather did not fall in step with arrivals. Arrivals per good-weather day fell sharply (see the weather chart).
• French preventions fell roughly in step with arrivals. The share of people stopped stayed near a third (see the preventions chart).
• Irregular crossings into the EU fell too, by almost a third in January–September. Most people who cross the Channel travel through the EU first.
• UK policy changed. Changes included enforcement against smugglers, returns to France and tighter asylum rules. Returns to France were about 4% of arrivals between September 2025 and June 2026.

The Migration Observatory concludes that UK policy and other factors probably both contributed, but that "it is difficult to isolate the effect of each factor".

The number of boats has fallen faster than the number of people. In the year to June 2026, 511 boats arrived, 33% fewer than the year before, and the average boat carried 65 people. That compares with 56 a year earlier and 7 in 2018. A single boat carried a record 230 people in August 2026. The Migration Observatory suggests that seizures of boats and engines may be one reason. It also reports a shift towards departures from Belgium in the first half of 2026.

From 2018 to 2025, people from six countries made up 65% of arrivals: Iran (16%), Afghanistan (14%), Iraq (10%), Eritrea (10%), Albania (8%) and Syria (8%).

Eritreans were the largest group in the year to June 2026, with 5,875 arrivals. Sudanese arrivals rose 11% to 4,191, and Somali arrivals rose 76% to 3,901. Syrian arrivals fell by 67% between 2024 and 2025, which the Home Office attributes to the fall of the Assad regime. Albanian arrivals rose and fell within 2022 (see the Albanian chart below).

From 2018 to 2025, detections by every other method fell: inside the UK (many of them people who came in lorries or on ferries), at UK ports, and on flights without proper documents. Small-boat arrivals rose by far more, so total detections more than tripled. Small boats have been the main detected method since 2020. The Home Office says the fall in other methods was "likely in part" due to COVID-19. By the year to June 2026, detections inside the UK were 77% below their 2019 peak.

The Migration Observatory says tighter security at French ports and the Channel Tunnel after 2014 is likely to have contributed to the rise of the boat route. Two limits apply. Detection rates differ by method, so the lines cannot be added up to a true total of people entering. And no official series covers the years before 2018, when the switch would have begun.

Home Office figures released under freedom of information show nearly 164,000 attempts at clandestine entry stopped at border controls in France and Belgium from 2010 to 2015. Most of them came in 2015, with 84,000 by Full Fact's count. Motor Transport reported 81,000 foiled attempts at the French ports in 2015 and more than 56,000 in 2016. The two 2015 figures come from different press reports and may cover different sets of controls. These are attempts, not people: one person can be stopped many times. They therefore cannot be compared with the post-2018 counts of people arriving.

Arrivals are lowest in mid-winter and highest from late spring to early autumn. The Home Office links this to crossing conditions. From January to September 2026, every month except February was below the same month of 2025.

Frontex recorded just over 91 000 irregular crossings into the EU in January–September 2026, almost a third fewer than a year earlier. Frontex attributes the fall to cooperation with partner countries and to preventive measures in departure states. The Migration Observatory cites Frontex figures showing EU crossings fell 26% in 2025 and another 37% in the first half of 2026. It expects this to reduce Channel crossings "with a lag", because most people who cross the Channel travel through the EU first. The UK's 2025 rise came while EU crossings were already falling, so the two do not move in lockstep.

From January to August 2026 the Met Office judged 110 days likely for crossings ("red days"). That is about a tenth fewer than the 123 in 2025, and more than in any year from 2020 to 2024. Over the same months, the average number of people arriving on a red day fell 38%, from 232 to 144, and was 43% below the 2022 peak of 251. The Home Office says 2026 had more favourable crossing conditions than almost any year since 2020. The Migration Observatory also judges weather "unlikely to have had a significant impact" over periods this long.

In the weeks to 18 October, the number of people the French authorities report stopping fell by nearly half compared with the same weeks of 2025, slightly more than arrivals fell. The share of people stopped stayed close to a third in both years. The share of events stopped rose, from a little over half to about three in five. Fewer, fuller boats would produce that pattern. The counts are French operational estimates. "Events" also include finds of equipment and arrests of facilitators, not only boats.

France changed its maritime law to allow boats to be intercepted at sea in limited circumstances, but it is unclear how often this has happened. No count is published.

The UK and Albania signed their returns agreement on 13 December 2022. Albania already accepted removals from the UK before then. Albanian arrivals had already fallen: from around 9,000 in July–September 2022 to around 1,000 in October, 166 in November and 28 in December. The fall was specific to Albanians: arrivals of other nationalities rose in the last quarter of 2022. The deal therefore cannot explain the initial collapse.

Albanian arrivals have stayed under a thousand a year since 2023, while arrivals from other countries recovered. Several things changed together after 2022, and the Migration Observatory says the deal "could be a major driver" of the continuing low numbers. The government prioritised Albanian asylum cases, and returns to Albania rose sharply. Albania accounts for 76% of all small-boat returns. The share of Albanian claims granted at first decision fell from almost half in 2021–2022 to 3% in 2024. No source separates the effect of the deal from faster processing, more refusals and more returns.

Over January–August, the UK's fall was larger than the EU total's and smaller than the falls on the Central Mediterranean, Western African and eastern land-border routes. Frontex's January–September release gives the Central Mediterranean down 56% and the Western African route down 45%. The Western Mediterranean rose 26%. The comparison shows that the UK's fall happened during a wider European one. It cannot show whether the two share a cause.

Claims peaked at around 103,000 in 2002, when most claimants came from Somalia, Afghanistan, Iraq and the former Yugoslavia, then ran between 22,000 and 46,000 a year until 2021. They rose from 2021, passed the 2002 peak and reached a rolling-year high of 109,662 in the year to September 2025. The Migration Observatory links the rise since 2022 to conflict in the Middle East and instability in the Horn of Africa and South Asia, and notes that not everyone from an unstable country qualifies for asylum.

85,891 people claimed asylum in the year to June 2026, 21% fewer than the year before. Claims across the EU fell faster in 2025, by 27%.

In the year to June 2026, 52% of people claiming asylum came by illegal entry routes, mostly small boats, and 41% had held a visa or other permission to be in the UK. Asylum claims are therefore not a count of illegal arrivals, and illegal arrivals are not a count of claims.

In every year from 2014 to 2024 the UK received fewer claims per head than the EU27 as a whole. The gap closed as UK claims rose and EU claims fell: in 2025 the UK received about 15 claims per 10,000 residents and the EU 1.5 per thousand, the same rate. Germany, which received the most claims per head of the large countries in 2015–2016 and again in 2023, fell below the UK in 2025.

The grant rate was low in the 2000s, rose through the 2010s and reached a record 76% in 2022, partly because more claims came from nationalities usually granted at high rates and partly because rates rose within nationalities. It fell to 42% in 2025 and 38% in the year to June 2026, against a rolling-year peak of 77% in the year to September 2022. The Home Office says grant rates move with the nationality mix, conditions in claimants' countries, and which cases are prioritised for decision.

Claims made from 28 June 2022 must show persecution "on the balance of probabilities", a higher bar than before. The Migration Observatory says the fall "may be linked" to this change; no published analysis measures how much it contributed.

Following each year's claims to their latest outcome, the grant rate for claims made in 2007–2022 ended on average 20 percentage points above the initial rate, and between 8 and 29 points above it in every year. For claims made in 2010–2019, the latest-outcome grant rate was between about half and three-quarters.

For claims made since 2023, the latest-outcome rate is barely above the initial rate. That reflects missing data, not a finding: the Home Office could not load appeals data for this release, and appeals now take over a year.

Withdrawn claims, which end without a decision, rose with the claims made from 2021. In 2023 the government widened the grounds on which a claim can be treated as withdrawn, though withdrawals were already rising. In the year to June 2026, 16,265 people had their claims withdrawn, 79% of them implicitly, usually for missing an interview or other step. A withdrawal is not a finding that the claimant was or was not a refugee.

39% of asylum and protection appeals determined in April–June 2026 were allowed. The allowed share rose from about a quarter in 2007/08 to about half in 2021/22–2022/23, and has fallen back since, to 39% in 2025/26. The appeals counted here were decided in that year and may concern claims refused years earlier.

The open asylum appeals caseload grew from 6,319 at the end of June 2022 to 87,450 at the end of March 2026, and an asylum appeal now takes 72 weeks on average to clear.

In the year to June 2026, 92% of Sudanese and 84% of Eritrean claims decided were granted, against 11% for Bangladeshis, 10% for Syrians and 1% for Indians.

The Syrian grant rate fell from 98% to 10% in a year after the fall of the Assad regime, and the Pakistani rate from 46% to 25%. Of the ten nationalities with most decisions in 2025, only Sudan was granted at a higher rate than in 2022.

Applying each nationality's 2022 grant rate to the 2025 mix of decisions gives a rate about seven points below 2022's. Applying 2025 rates to the 2022 mix gives a rate close to 2025's. So the shift towards lower-grant nationalities such as Pakistanis, Bangladeshis and Indians explains a minority of the fall; lower rates for the same nationalities, notably Afghans, Syrians, Turks and Vietnamese, explain most of it. This calculation cannot say why rates fell within a nationality: a higher standard of proof, changed country conditions and a different mix of cases decided are all candidates.

Of 178,845 people who arrived by small boat in 2018–2025 and claimed asylum, 79,589 had been granted at initial decision and 53,305 refused, a grant rate of 60% on cases. Rates vary by nationality: 86% for Eritreans who arrived in 2025, 34% for Afghans.

For people who arrived in 2023 and 2024, refusals slightly outnumbered grants at initial decision; for 2020–2022 and 2025 arrivals, grants were the clear majority. Across all arrivals, 33,413 claims, or 18.7%, were withdrawn or closed administratively without a decision on the merits. Withdrawals were concentrated among 2022 arrivals. These figures exclude appeals.

39% of first-instance decisions across the EU were positive in 2025, and 21% of final decisions after appeal or review. The UK rate on the same people-based measure was similar. In 2024 the UK's rate was 47% against the EU's 51%; the Migration Observatory notes that overall rates mostly reflect which nationalities claim where.

The UK has granted higher rates than the EU for most of its main nationalities since 2018, but grants Afghans far less often: 51% against 81% in 2024, and in 2025 the EU granted 73% of Afghans. In 2025 the UK also granted Somalis and Syrians at lower rates than the EU, and Iranians, Sudanese and Vietnamese at higher ones.

40,168 people were waiting for an initial decision at the end of June 2026, 77% fewer than at the June 2023 peak of 175,457 and the lowest number since June 2019. The queue is still longer than it was through the 2010s: it grew from 7,240 in December 2010 to 35,855 in December 2018.

Hotel use fell by the same order: 16,021 supported asylum seekers were in hotels at the end of June 2026, 71% below the peak of 56,018 in September 2023 (see the accommodation charts below).

As the first-decision queue shrank, the tribunal's queue grew. Open asylum and protection appeals at the First-tier Tribunal rose from 6,319 at the end of June 2022 to 90,341 at the end of June 2026, more than fourteen times as many. By June 2026 the appeals line was more than twice the first-decision line, though the two count different things (see Subtleties).

The National Audit Office traces this to the push to clear the legacy backlog: the Home Office made more than four times as many initial decisions in 2023 as in 2022, and "this shifted pressure into the appeals stage where constraints on legal and judicial capacity created another backlog". The tribunal dates the steep rise in new asylum appeals to late 2023 and the same backlog clearance, and the Home Office's accounts attribute it to "higher volumes of initial asylum decisions flowing into the tribunal system".

Most refusals are appealed: the Home Office's evidence to the NAO is that 68% of people refused at initial decision in 2024-25 lodged an appeal. An asylum or protection appeal took a mean of 72 weeks to clear in April to June 2026. New appeals have started to fall in the latest two quarters, but the open caseload was still rising in June 2026.

• 2021 to 2023: more claims, few decisions. Claims rose between the years to June 2021 and June 2023 while initial decisions stayed comparatively low, so the queue grew.
• March 2023 to mid-2024: decisions on hold. Many claims made on or after 7 March 2023 could not be decided under the Illegal Migration Act, until the government removed its retrospective application in July 2024. The queue rose again in late 2024 before falling.
• From late 2024: more decisions. The Home Office attributes the fall to more decision makers and higher productivity; the queue has fallen every quarter since decision making fully resumed in October to December 2024.

39% of asylum and protection appeals determined at a hearing or on the papers in April to June 2026 were allowed. That share excludes appeals that ended without a judge's decision: across all appeal types, 29% of the quarter's disposals were withdrawn and 17% struck out for an unpaid fee.

93,293 people were on asylum support at the end of June 2026, 25% below the September 2023 peak of 123,758, a much smaller fall than the 77% in the first-decision queue. The Home Office notes that with 40,168 people awaiting an initial decision, most people on support are in other parts of the system, such as awaiting an appeal or refused. 5% (4,987) were on Section 4 support, which is for people whose claim has been finally refused but who cannot yet leave.

The number of people in dispersal housing, the longer-term homes the contracts provide, has risen in most years since 2014 and did not fall with the queue. At the end of June 2026, 73,068 people (78%) were in non-hotel accommodation, 3% more than a year earlier. Part of the fall in hotel use is people moving to other accommodation rather than leaving support.

16,021 people were in hotels at the end of June 2026, 17% of everyone on support, half the number a year earlier and 71% below the peak of 56,018. The government said in August 2026 that fewer than 160 asylum hotels remained in use, against 400 at the 2023 peak. Those are hotel counts from a ministerial announcement, not from the statistics.

At the end of June 2026, 744 people were at Wethersfield and 418 at Crowborough, the two former military sites in use. They sit in the second line of this chart, with initial and other contingency accommodation.

The Home Office spent around £4.0 billion on asylum support in 2024-25, down from £4.7 billion in 2023-24. It had spent around £739 million in 2019-20 and £3.6 billion in 2022-23. Counting casework, appeals, detention and unaccompanied children as well, the NAO puts Home Office and Ministry of Justice spending on asylum at around £4.9 billion in 2024-25, without local authorities' costs. The government expects reforms to cut asylum costs by at least £1 billion a year by 2028-29 against 2024-25.

Hotels took £2.3 billion of the £3.6 billion spent in 2022-23. Hotel spending was £3.0 billion in 2023-24 and £2.1 billion in 2024-25, and the Home Office's accounts say asylum support spending rose "due to the increased use of hotel accommodation, peaking in 2023-24, with year-on-year reductions in 2024-25 and 2025-26".

The asylum business group was 29% of the Home Office's Resource DEL outturn in 2023-24 and 23% in 2025-26 (asylum line divided by the total, from the core table). The total is net of visa and passport fee income, which makes the asylum share larger than it would be against gross spending, and the asylum line includes resettlement, such as £222.1 million for Afghan resettlement in 2025-26. Plans to 2028-29 have the line falling every year.

Accommodation and support was £3,406 million of the NAO's estimate for 2024-25. Appeals cost about £102 million and asylum detention and removals about £46 million, the smallest items in the figure.

In the first seven months of 2024-25, people in hotels were around 35% of everyone in asylum accommodation and around 76% of the contracts' cost: £1.3 billion out of an estimated £1.7 billion. Hotels cost from around £127 to £148 per person per night between April 2023 and January 2024, and suppliers' data suggest hotels may be more profitable for them than other accommodation.

The Home Office estimated the accommodation contracts at £4.5 billion over ten years when it let them; the estimate is now £15.3 billion over the same period.

In March 2024 the NAO reported the Home Office's own analysis that, including sunk costs, its four large sites would cost about £46 million more than hotels. By May 2025 the Home Office no longer planned to use large sites other than Wethersfield, having found that "costs increased" and that housing many people near small communities "is difficult to manage". In 2026 it reversed course: Crowborough opened in January, and the three proposed MOD sites could house around 3,750 people. Cherwell District Council gives the total as 3,700. No cost estimate for the new sites appears in the announcement, the factsheets or the council statement cited here.

In the NAO's comparison countries, Germany houses asylum seekers in initial reception centres for up to 18 months before collective centres, and most people entitled to asylum accommodation in the Netherlands in early 2025 stayed in emergency sites such as halls, ships and tents. The NAO compared how these systems work, not what they cost per person, so there is no like-for-like cost comparison with the UK here.

Enforced returns drifted down from 2004 to 2019 and voluntary returns fell from 2016, before both collapsed during the pandemic. Since 2021 enforced returns have risen each year, to 9,669 in the year to June 2026, and voluntary returns to 30,936. The Migration Observatory puts 2025 returns at the highest since 2016 but still slightly below a decade before.

The Migration Observatory attributes the 2010s decline to a combination of resource constraints, more legal challenges, administrative delays and fewer referrals to Immigration Enforcement, and the recovery to bigger budgets and higher immigration. Enforcement spending fell 19% in real terms between 2014/15 and 2020/21. It says limited data prevent a definitive assessment of the legal-challenge share.

An enforced return cost an average of £48,800 in 2024/25, against £4,300 for a voluntary one. Port returns — people refused entry who leave again — are counted separately: 18,646 in the year to June 2026. All these figures count returns, not people: someone returned twice in a period is counted twice.

Countries without a formal UK returns agreement accounted for 49% of returns in 2025, and the Migration Observatory cites studies finding agreements had little effect on return rates in the EU. Failing to obtain travel documents was the most common reason officials decided not to detain someone liable to removal, in a 2014/15 inspection.

Foreign offender returns averaged around 5,500 a year in 2010–2019, peaked at 6,437 in 2016, fell below 3,000 a year in 2020–2021, and in the year to June 2026 were 7% below the peak. The Home Office attributes the fall after 2017 partly to a 2017 Supreme Court judgment that stopped removal before appeals were heard and a 2019 legal challenge that suspended charter flights, as well as the pandemic.

12,568 asylum-related returns in the year to June 2026 were the most since the series began in 2010, and 31% of all enforced and voluntary returns, up from 15% in 2021. Of everyone who has arrived by small boat since 2018, 9,694 had been returned by June 2026.

Of asylum claims made in 2010–2022 and refused by the end of 2025, around half had led to a recorded return. The share was around 72% for 2010 claims, 33% for 2018 claims and 48% for 2022 claims. The chart's own measure, which also counts withdrawn claims, shows the same shape: above half for claims made before 2014, around a fifth to a quarter for 2016–2021.

The government said 58,000 people had their asylum claim refused between June 2024 and June 2025, while fewer than 11,000 people were removed; the two counts are different groups of people, not a removal rate. The Bill's impact assessment sets 446,000 claims in 2021–2025 against 33,000 asylum-related returns.

193,000 people who claimed asylum in 2010–2024 and were refused or withdrew had not been recorded leaving by January 2026. The Migration Observatory treats this as an upper bound, because some were later granted status or left unrecorded, including via Ireland.

Before 2015 most people leaving detention were returned. In every year since, more have been bailed than returned, though the gap has narrowed. Returns from detention have since recovered: 46% of people leaving detention in the year to June 2026 were returned, against 25% in the year to June 2023. The Home Office says people who claim asylum or raise other claims or appeals may be bailed while the claim is decided.

Entries fell from 32,447 in 2015 to 24,480 in 2019, partly after the 2016 Shaw review. 2,152 people were held at the end of June 2026, the most since June 2018. In 2019 the National Audit Office found the Home Office released 14,900 (62%) of the detainees it intended to remove.

In 2019 people facing removal from detention after immigration offences raised at least one issue in 73% of detentions, and 94% of those were released. The commonest issue was an asylum claim, in 47% of detentions, followed by medical reports and modern slavery referrals. Human-rights and other rights-based applications fell from 10% to 5% of detentions between 2017 and 2019, and judicial reviews from 20% to 9%.

Of issues raised in detentions ending in 2017, 12% led to a positive decision or concession, 83% did not, and 5% were still open. When no issue was raised, 67% of detentions ended in return.

In January to September 2022, 86% of people who raised rights-based applications in detention were released, and more than 2 in 5 people detained for removal were referred as possible victims of modern slavery. The fuller update has not been published.

The Home Office set 25,700 removal directions in 2019 and cancelled 13,500 (52%), against 11% of removals failing in 2013. Immigration Enforcement blamed late or spurious asylum, medical and modern slavery claims; the National Audit Office saw no systematic analysis of why claims were increasing. It listed other barriers: conflict at home, a long-established life in the UK (human-rights grounds), missing identity documents, and no returns agreement — but gave no breakdown by reason.

The two flights planned 144 and 163 removals and removed 42 and 32. Fresh asylum claims after removal directions were set were the commonest reason for failure (55), then judicial reviews (35) and further representations (26). Caseworkers believed many late submissions had no merit and aimed to frustrate removal. About a fifth of the failures in the top five reasons were people the Home Office did not find, or did not visit for lack of resources.

Of 24,289 planned foreign-offender removals in 2014/15–2016/17, 16,517 (68%) succeeded and 7,772 (32%) failed. In a review of 1,653 non-detained cases, almost 80% had at least one barrier to removal, including outstanding claims or missing travel documents; only 194 were judged removable.

Appeals allowed on human-rights grounds only rose to 337 in 2012/13, fell by roughly three-quarters to 92 in 2015/16 after the Immigration Act 2014, then climbed back to 285 in 2017/18. The 2014 Act raised the bar for offenders' appeals while leaving judges some discretion. The rebound coincides with the 2017 Supreme Court judgment that paused "deport first, appeal later", so the data cannot separate the two.

From 2015/16 to 2020/21 about 1,000 foreign offenders won appeals on ECHR grounds alone, against 31,400 removed: 3%. The Bonavero Institute's count for April 2016 to June 2021 is 922 against 26,091, or 3.5%, and around 2.5% for Article 8 alone. A sample found 70% of human-rights wins rested solely on Article 8.

Across all human-rights immigration appeals, not only deportations, judges allowed 61,000 of 115,000 human-rights appeals from 2015/16 to 2024/25 (53%). These are mostly not removal cases, so they do not measure blocked removals, but they show the ECHR's reach in immigration decisions is wider than offender deportations.

21,294 foreign offenders liable to deportation were living in the community at the end of June 2026, 12% more than a year earlier, with most of the latest rise after the Sentencing Act 2026 widened the duty to deport. The Migration Observatory reads the earlier growth, 36% from the end of 2022 to March 2026, as a sign that removals have not kept pace with prison releases. The government does not publish what share of offenders are deported.

The Conservative "Removals Force" pledge of 150,000 removals a year is about 15 times the 9,669 enforced returns in the year to June 2026, and well above the highest year of enforced returns in the series (2004). It is also more than all 27 EU states returned to non-EU countries in 2025.

The pledge comes with £1.6 billion. At the 2024/25 average of £48,800 per enforced return, that would pay for about a fifth of 150,000 enforced returns; at £4,300 per voluntary return it would cover the target several times over. For scale, 2,152 people were held at the end of June 2026.

The US deported about 980,000 people in the first 18 months of the second Trump term, but 300,000 of those followed arrests inside the country; most were border cases.

By 30 June 2026, 1,087 people had been returned to France and 1,117 brought to the UK under the pilot. 27,920 people crossed by small boat over the same period, so about 4% were sent back. Monthly returns rose through spring 2026. The Home Office said on 30 September 2026 that the pilot was no longer processing new cases. The Home Secretary put the cost at about £56,000 per return.

The Home Office makes no official estimate of the population without lawful status. The two latest estimates are for 2017, and Pew revised its range down in March 2025. Both use the residual method: the foreign-born population minus an estimate of those here lawfully.

No published source counts removals stopped by the ECHR. The Home Office does not hold data on people granted status after being told they were liable to removal. The proxies above each answer a narrower question, and both may understate the ECHR's weight.

The National Audit Office named conflict at home, missing identity documents and countries without return agreements alongside human-rights grounds. A fifth of failures on the Jamaica charters were people the Home Office did not find or did not visit.

Boat arrivals were low for most of the period from 1976, with a smaller peak around 2001, then rose from 2009 to the two highest years in the series, 2012 and 2013. More than 24,000 people arrived between 13 August 2012, when offshore processing restarted, and 18 July 2013.

During the first phase of offshore processing, which still allowed eventual settlement in Australia, the Kaldor Centre found "no noticeable change in the rate of arrivals". After the government announced on 19 July 2013 that no one arriving by boat would ever settle in Australia, more than 1,500 people on at least 21 boats arrived in the first 16 days.

Operation Sovereign Borders began in September 2013, and the first recorded turnback was on 19 December 2013. Arrivals fell to 160 people in 2014 and none in 2015 or 2016. By October 2016, 740 people on 29 vessels had been returned to their country of departure since December 2013.

The no-settlement rule (July 2013), the start of Operation Sovereign Borders (September 2013) and the first turnback (December 2013) all came in the same year, so annual figures alone cannot separate them. Monthly figures were not found (see Data still needed).

The Kaldor Centre, which opposes offshore processing, reads the fall in arrivals as following the start of maritime interceptions, and attributes the end of deaths at sea to turnbacks and takebacks rather than offshore processing. It points out that Australia stopped sending new arrivals offshore in 2014 and returned boats instead.

The Australian government described offshore processing as part of a "broad chain of measures end to end", and as minister Peter Dutton argued that letting people move from Manus to Australia would bring the boats back. Both readings fit the annual data.

Operation Sovereign Borders defines a turnback as removing a vessel from Australian waters with passengers returned to their country of departure, and a takeback as a transfer to that country's authorities "where Australia works with a country of departure".

Australia has ratified the 1951 Refugee Convention and its 1967 Protocol. The Kaldor Centre argues that offshore processing breached those obligations. Australia has no bill or charter of rights, so the policy could not be challenged in Australian courts on the ground that it broke international law.

Of the 1,241 people on boats from September to December 2013, 1,107 were brought to Australia or sent offshore and 134 returned. In 2014, 157 were brought in and 293 returned. From 2015 to 2019, all 385 people on 23 boats were returned (217 + 51 + 60 + 24 + 33 people; 9 + 6 + 3 + 2 + 3 boats). These are Home Affairs figures as reproduced by the Kaldor Centre.

The Refugee Council of Australia reports that transfers to Nauru resumed in late 2023, for the first time since 2014, with around 105 people held there in February 2026.

Two sets of government figures compiled by the Kaldor Centre put the cost at an average of at least A$1 billion a year, peaking at A$1.49 billion in 2017-18. The Refugee Council of Australia puts total spending since 2012 at more than A$13 billion.

The number held on Nauru and Papua New Guinea fell from a peak of 2,450 in April 2014 to 239 in May 2021, while annual spending stayed in the hundreds of millions. The Kaldor Centre reads this as evidence of large fixed costs.

Of approximately 4,180 people transferred from 2012 to 2014, 2,043 were back in Australia by January 2021, 921 had been resettled in a third country and 935 had returned to their country of origin; 18 had died.

The Refugee Council of Australia's count to December 2025 covers 4,359 people transferred, including those sent since 2023: 1,547 resettled in third countries, 978 returned or removed, and 21 dead.

The Home Office spent £715 million on the Rwanda partnership and the Illegal Migration Act up to June 2024, including £290 million paid to Rwanda. Four volunteers and no asylum seekers were sent to Rwanda. Ending the partnership avoided over £220 million in scheduled payments.

The £715 million covers the Illegal Migration Act as well as the partnership, so it is not all the cost of relocating four people. The payments to Rwanda were mostly fixed sums due whatever the number relocated, and the other categories are largely fixed costs; per person, the agreement also provided up to £150,874 for processing and support over five years.

The National Audit Office did not judge value for money, because that rests on whether the partnership deterred journeys. With no asylum seekers relocated, there is no evidence on that either way.

Denmark had 1,835 first-time asylum applicants in 2025, 3.1 per 10,000 residents, against 14.9 across the EU27. The UK received 15 per 10,000 in 2025 on the Migration Observatory's count of all applicants.

Denmark's first-time applicants fell from 20,825 in 2015 to 6,055 in 2016, 29% of the year before. Sweden's fell further, from 156,115 to 22,335, 14% of the year before. Sweden also moved from permanent to temporary residence permits in 2016. The EU27 total in 2016 was still 96% of 2015 and fell sharply only in 2017, so the Danish and Swedish falls came about a year ahead of the wider European one.

Since 2015 Denmark has made refugee status temporary with review every two years, restricted family reunion, and lengthened the wait for permanent settlement to eight years, with integration and employment conditions. Its applicants peaked in 2015, the year temporary status began, during the Europe-wide rise. The data cannot attribute the later fall to any one of these measures.

The Home Secretary's March 2026 statement said Denmark had cut claims to a 40-year low while UK claims rose 13% and EU claims fell 22% in the year to September 2025. Eurostat's series starts in 2008. Within it, Denmark's 2025 total of 1,835 is the second-lowest year, above 2020 (1,420), the first year of the pandemic.

In 2025, 36% of Denmark's first-instance decisions were positive, against 39% across the EU27. In 2016 the figures were 68% and 62%. Denmark's low intake is not explained by a lower share of positive decisions at first instance.

On first-time applicants per 10,000 residents in 2025, Denmark sits in the bottom third of the 31 EU and EFTA countries, below Sweden and well below Germany, which is close to the EU27 average.

Studies of Denmark's 2002 reform and Germany's 2016 reform found no employment gain from making permanent residence harder to get. Denmark's 2007 reform reduced work hours, driven by refugees who had been doing worst before it. Sweden's 2016 switch to temporary permits had no effect on employment in the first year.

Jutvik and Robinson found that refugees on temporary permits had higher incomes and less unemployment in the short term than those given permanent residence, but spent less time in education.

All five findings come from Denmark, Germany and Sweden, and they measure employment rather than the number of people who claim asylum. No evaluation of the UK's 30-month protection, which began on 2 March 2026, was found.

The one cross-country study of destinations finds that shorter work bans go with more applications, but the association is about a quarter the size of the association with earlier applications from the same country (0.18% against 0.66% for each one per cent change). Its authors conclude that work bans "are not a strong deterrence". It is a working paper on EU countries, not the UK.

Studies across Europe, Germany and Switzerland find that delaying work or decisions lowers refugees' employment for years afterwards. In 2021 the Migration Advisory Committee urged the government to review the UK ban, citing "clear evidence of the harm" and "little evidence" of significant benefits.

In Parliament the Home Secretary said in November 2025 that a right to work while waiting "would be a huge pull factor", and in September 2025 a Home Office minister said cutting the wait to three months "could act as a pull factor". No published Home Office estimate of the size of that effect was found (see Data still needed).

The March 2026 statement calls the reset "inspired by Denmark’s success" and aims to "reduce the pull factors driving illegal migration". The Danish evidence above shows a low and falling intake, alongside a Swedish fall of similar timing and a grant rate near the EU's.

The Immigration and Asylum Bill was introduced on 30 June 2026 and was still progressing through Parliament in July 2026. The 30-month rule already applies through the Immigration Rules.

From 2010 to 2020, out-of-country grants ran between about five and thirteen thousand a year, starting at 5,603 in 2010. They jumped when the BN(O) route opened on 31 January 2021 and the Ukraine schemes in March 2022, peaking at 273,413 in 2022 . Ukraine grants peaked at 209,706 in 2022 and BN(O) grants at 86,439 in the year to March 2022, then both fell . The Ukraine Family Scheme closed to new applications on 19 February 2024 . Out-of-country grants fell to 48,814 in 2025, and to 38,267 in the year to June 2026, 29% fewer than the year before .

The Home Office headline counts 124,930 grants on these routes in the year to June 2026, but 69% went to people already in the UK extending their leave . This chart leaves extensions out. Visa grants also overstate arrivals: Ukraine arrivals peaked at 110,000 in 2022, against 209,706 visas granted .

Before 2021, resettlement was mostly Syrians: 22,157 people came through the Vulnerable Persons and Vulnerable Children's schemes between 2014 and 2021 . It stopped in mid-2020 for the pandemic . The Afghan Resettlement Programme has resettled 39,058 people since 2021 , but its schemes closed to new applicants or referrals in July 2025, with existing applications still being processed . In the year to June 2026, 3,840 people were resettled, 48% fewer than a year earlier; 87% (3,352) were on the Afghan programme and 13% (488) on the UK Resettlement Scheme and the Mandate Scheme .

Refugee family reunion lets the partner and children (aged 17 and under) of a refugee in the UK join them, if they were part of the family before the refugee fled . It ran at 4,113 to 7,456 grants a year from 2010 to 2022, then rose to 9,320 in 2023, 19,258 in 2024 and 18,577 in 2025 . The Home Office links the rise to the large number of people granted refugee status in the same period . The route was paused on 4 September 2025, as a "temporary pause" pending a review of family rules . The Immigration Rules still describe it as "closed to new applications pending a review" . Grants in the year to June 2026 were 12,888, 38% lower, and still include applications made before the pause .

The UK Resettlement Scheme began in February 2021 and has resettled 4,589 refugees, from 1,269 in its first year to 537 in 2025 . In September 2025 the Home Office confirmed "there is currently no 2025/2026 quota for new referrals to the UKRS", with arrivals continuing only for people already accepted .

In 2018 to 2020 there were 10,608 small-boat arrivals in total; from 2021 the yearly figure has been between 29,000 and 46,000 . Out-of-country safe-route grants were above that every year from 2021, peaking at 273,413 in 2022 . By 2025 the two were close, as the Ukraine and BN(O) schemes shrank. Most of the grants went to nationalities that were not crossing by boat (see the next chart).

From 2018 to June 2026, 12 Ukrainians and no Hong Kong nationals were recorded arriving by small boat, against 290,044 Ukraine visas and 187,552 visas on the BN(O) route . Both schemes were for people from one place: the Ukraine Sponsorship Scheme for Ukrainians and their families with a UK sponsor, and BN(O) for BN(O) passport holders resident in Hong Kong .

For Eritreans, Sudanese, Iranians, Ethiopians and Yemenis, more than nine in ten safe-route grants from 2018 to June 2026 were family reunion visas, and for Somalis about two-thirds; resettlement was small and the Ukraine and BN(O) schemes did not apply . Family reunion depends on someone in the family already holding refugee status in the UK , and it is the route that was paused in September 2025 . Eritreans, Sudanese, Afghans, Somalis and Iranians were the five largest small-boat nationalities in the year to June 2026, two-thirds of all arrivals .

Afghans had the largest resettlement route (the Afghan Resettlement Programme, 39,058 people since 2021) and were still among the largest small-boat nationalities over the period . Syrians had both resettlement and family reunion at scale, and also crossed by boat in large numbers. A route did not stop crossings where the route was small relative to the number of people who wanted to come, or where it was open only to some of them; these data cannot say which.

Albanians, Iraqis, Vietnamese and Turkish nationals crossed by boat in large numbers and received far fewer safe-route grants than boat arrivals; Albanians received almost none . The data do not say whether they would have qualified for a route.

In the year to June 2026 small-boat arrivals were 33,374, 23% fewer than the year before . Eritreans were the largest group at 5,875; Sudanese arrivals rose 11% to 4,191 and Somali arrivals 76% to 3,901 . For each of these, and for Iranians, safe-route grants were a fraction of boat arrivals, and most were family reunion visas on applications made before the pause . Afghans, Syrians and Yemenis received more safe-route grants than they had boat arrivals, through resettlement (Afghans) and family reunion. Afghanistan, Yemen, Sudan, Iran and Syria were the top family reunion nationalities, at 17%, 13%, 11%, 11% and 9% of grants .

On 28 September 2026 the Home Secretary reopened the UK Resettlement Scheme to new referrals, with around 1 in 4 expected to be Afghan women and girls and Palestinian refugees . Three new sponsorship routes (community, student and work) are to follow, with first arrivals on the community and student routes expected in autumn 2027 . People on these routes get a five-year path to settlement, which the government describes as "by design, more generous" than for those who arrive by small boat .

The government expects arrivals across resettlement and the three new routes "to number in the hundreds at first", growing "into the low thousands as levels of illegal migration fall", under a cap set each year . For scale, there were 33,374 small-boat arrivals and 12,888 family reunion grants in the year to June 2026 . Family reunion remains closed .

Since 28 June 2022, a refugee who stopped in another country on the way to the UK cannot rely on the Article 31 defence unless they show they could not reasonably have sought protection there. McConnachie's study of the case law says the effect is "to exclude almost all refugees who enter or leave the UK from protection under article 31", because few reach the UK without passing through another state. On this point UK statute already works as a first-safe-country rule.

UNHCR's 2024 guidelines follow Adimi: refugees have "some element of choice", and a short stopover does not break "coming directly". UNHCR describes the UK's statutory rule as a misapplication of the Convention, so UK law and UNHCR's reading of the same treaty now diverge.

Article 31 forbids penalties for illegal entry. It does not by itself decide whether a state must consider an asylum claim or where a person may be sent; sending a refugee somewhere they would be in danger is barred separately, by Article 33(1). The EU applies its own "first country of asylum" concept through the Dublin rules.

Wolfson advised that the UK could lawfully leave the ECHR, that legislating inside it would probably be litigated to failure, and that leaving the Refugee Convention is not needed to meet the Conservative policy goals on immigration: "a stringent regime could be pursued outside of the ECHR, while remaining within the Refugee Convention".

Wolfson's conclusion is that the ECHR "places significant legal constraints" on immigration policy. The review does not estimate how many removals those constraints prevent.

Under Article 692(2), if Part Three is ended because the UK denounces the ECHR, it stops on the date the denunciation takes effect. Ending it is the EU's choice, not automatic. Either side can also end Part Three for any reason, in which case it stops on the first day of the ninth month after notice; Wolfson describes this general right as twelve months' notice. Wolfson calls the consequences of withdrawal "more likely to be political than legal"; CAJ calls them "serious consequences" for policing and criminal-justice cooperation.

The two sides read the same commitment differently: as a promise to make ECHR rights part of Northern Ireland law, or as a promise that also requires the UK to stay a party to the Convention. No published UK government legal assessment was found.

The Refugee Convention's Article 33(1) forbids returning a refugee "in any manner whatsoever" to where their life or freedom would be threatened. In UK practice, most claims of risk on return are asylum claims under the Refugee Convention, not Article 3 ECHR claims.

The declaration is a political statement by governments, reported on adoption as "non-binding". It does not amend the Convention. Its Article 8 paragraph restates the Court's own case law ("According to the case law of the Court…").

All 46 states adopted it, but the minutes record Spain dissociating itself from the references to "return hubs".

The impact assessment models about 11,700 more Article 8 refusals a year after appeal, but says "it is not a target or forecast" and that a fixed target would not be appropriate. It estimates that 55% of people refused under Article 8 may not leave the UK, and expects 3,600 more a year to be removed or to claim asylum.

The Public Law Project says there is "no evidence that the Bill will achieve its policy objectives". The Commons Library notes the tension between statutory rules on Article 8 and the fact that the Strasbourg Court decides what Article 8 means.

48% backed withdrawal in 2013 and 41% in July 2014, against 35% and 38% for remaining. By 2016 the order had flipped, 42% to 35%. YouGov recorded 51% to 27% in June 2025 and 46% to 29% in October 2025, with 24% unsure. A Savanta poll for Amnesty International, reported in November 2025, found 48% for staying and 26% for withdrawing.

In June 2025, 54% of Conservative voters and 72% of Reform voters favoured leaving. In October 2025, 72% of Reform voters and 44% of Conservative voters did, against 82% of Labour voters who wanted to remain.

A question that does not name the ECHR but promises to "promptly deport illegal migrants" drew 54% support, while offering middle options cut outright support for leaving to 16%.

In September 2023, 38% chose immediate removal and 43% case-by-case decisions; 63% of 2019 Conservative voters chose removal and 71% of 2019 Labour voters chose case by case.

In August 2025, 45% backed requiring large numbers of recent migrants to leave. Counted across all adults, that came to 38% who included small-boat asylum seekers and 16% who included asylum seekers who came through the legal process.