GEOPOLITICA
418 to 218 — the EU rewrites its expulsion rules: 29 states, centres outside Europe and a year of preparation

The European Union has replaced its 2008 expulsion rules with a directly applicable regulation, adopted by Parliament and the Council, that allows detention of up to 30 months and transfers to return centres outside Europe.
What is at stake is not a change of principle: the old directive also required states to issue return decisions. What changes is the instrument. A directive was transposed by each state in its own way, while a regulation applies identically, from the same day, from Lisbon to Tallinn. On top of that comes a common form, the “European Return Order”, which lets one state's decision be enforced by another without the procedure starting again from scratch.
The testable thesis of this analysis: the vote closes a legal gap, not a practical one. The rate at which people actually leave depends above all on countries of origin taking back their citizens, and no European law changes that. The part with immediate effect, centres outside the EU, now has a legal basis but not yet an address.
What was voted, in order
The Commission proposed the regulation on 11 March 2025 as the missing piece of the Pact on Migration and Asylum, which has applied since June 2026. Commissioner Magnus Brunner summed up the problem in one sentence at the time: “One out of five people who are told to leave the EU actually leave the EU, and that is not acceptable.” Parliament adopted its negotiating mandate on 26 March 2026 by 389 votes to 206, and on 1 June negotiators from the two institutions reached a political agreement.
The plenary adopted the compromise text on 17 June with 418 votes in favour, 218 against and 30 abstentions, after a motion to reject had failed by 198 to 453, according to the roll-call record published by HowTheyVote.eu. Malik Azmani, Renew's lead negotiator, called it the “final missing piece” of Europe's migration system: “After almost 20 years of standstill, Europe finally has effective return measures,” according to Brussels Signal.
Interior ministers gave final approval on 1 October in Luxembourg. Twenty-three states voted in favour, Belgium abstained, Spain voted against, and Denmark and Ireland do not take part under their justice and home affairs opt-outs, according to Agence Europe. The regulation enters into force on the day after its publication in the Official Journal. As of 12 October we had not yet found it published.
What the final text actually changes
We read the adopted text (document PE-CONS 40/26). The gap between the figures quoted in the press, 24 or 30 months of detention, is resolved in Article 30: detention may not exceed 12 months in a member state, may be extended up to 24 months if return is delayed by the person or by the country of origin, and, after 24 months, by at most six more months in total if the person poses a security risk. Both figures are right, at different thresholds.
| Element | 2008 directive | Regulation adopted in 2026 |
|---|---|---|
| Legal form | Directive, transposed differently in each state | Regulation, applied directly and identically |
| Voluntary departure period | 7–30 days | At most 30 days, with justified exceptions |
| Detention | 6 months, extendable to 18 | 12 months, extendable to 24, plus at most 6 for security risks |
| Entry ban | As a rule, at most 5 years | At most 10 years; up to 20 years in security cases |
| Another state's decision | Optional recognition, rarely used | European Return Order, entered in the Schengen Information System |
| Country of return | Country of origin or transit | Also a third country with which an agreement exists: “return hubs” |
| Suspensive effect of appeals | Left to the states | Not automatic; a court may suspend removal on request |
A person without a right to stay now has explicit obligations: to cooperate with identification, to hand over travel documents and biometric data, and not to abscond. To “prepare or ensure an effective return”, authorities may enter the person's home or other relevant premises and seize phones and documents, according to EUobserver. Belgium is already preparing a law allowing forced entry between 5am and 9pm.
Return hubs have an article of their own, Article 14. A transfer requires an agreement with the third country, and such an agreement may only be concluded with a state that respects international human rights standards and the principle of non-refoulement. The agreement must set out, among other things, the consequences of a breach and an independent monitoring mechanism. Unaccompanied minors may not be sent to such centres. Families with children may, even though the Commission's proposal had excluded them, notes ECRE.
How many countries are affected: three different counts
The question “how many countries” has three answers, depending on whether we count the states applying the rule, the states voting on it, or the states whose citizens are returned.
States applying the regulation: 29, possibly 30. The regulation binds 25 of the 27 member states. Ireland does not take part in this part of the Schengen acquis. Denmark is not bound by it, but because the text builds on the Schengen acquis it has six months from the Council's decision to decide whether to implement it in national law, so until early April 2027. Beyond the EU, the regulation also applies to the four Schengen associated states: Norway, Iceland, Switzerland and Liechtenstein.
States that voted for it: 23. Of the 25 states that voted in the Council, 23 were in favour, Romania among them. Spain's vote is the most interesting: Madrid does not reject faster returns, but it cites “serious doubts” about the legality and proportionality of centres in third countries and fears for its relations with partners across the Mediterranean, according to the Spanish government and Euronews.
Countries of origin: practically all, but concentrated on a few. In 2025 EU states issued 486,705 orders to leave, according to the Eurostat series updated on 30 September. The top five citizenships, Algeria, Morocco, Türkiye, Syria and Afghanistan, account for 137,240 orders, or 28%. The top ten reach 210,515, or 43%. Algerians, with 45,265 orders, are almost twice as many as the next nationality on the list, and France alone issues 137,550 orders, more than a quarter of the total.
| Citizenship | Orders to leave in the EU, 2025 | On the EU list of safe countries of origin |
|---|---|---|
| Algeria | 45,265 | No |
| Morocco | 28,880 | Yes |
| Türkiye | 24,780 | Candidate country: yes in principle, with the exceptions provided |
| Syria | 20,055 | No |
| Afghanistan | 18,260 | No |
| India | 16,475 | Yes |
| Tunisia | 16,240 | Yes |
| Georgia | 14,095 | Candidate country: yes in principle, with the exceptions provided |
| Senegal | 13,280 | No |
| Bangladesh | 13,185 | Yes |
The third column links the June vote to the one on 10 February, when Parliament approved the safe third country concept by 396 votes to 226 and the EU list of safe countries of origin: Bangladesh, Colombia, Egypt, India, Kosovo, Morocco and Tunisia, plus the EU accession candidates, according to Al Jazeera. The two texts work together. Asylum claims from citizens of listed countries are examined in accelerated procedures, and the Return Regulation takes over the rejected cases.
The causal chain: why now
The cause. The old directive produced decisions, not departures. In 2025 the ratio of effective returns to third countries to orders to leave was 28.6% across the EU, but with huge differences: 11% in France and Spain, 53% in Germany, 61% in Sweden. One state could not enforce another's decision, so if the person crossed an internal Schengen border, the state where they arrived started the procedure from scratch.
The political mechanism. The 2024 European elections shifted Parliament's centre of gravity to the right. The breakdown of the 17 June vote shows the coalition: EPP 166 for and 1 against, Patriots for Europe 79 to 0, ECR 78 to 0, Europe of Sovereign Nations 26 to 0. Renew split (37 for, 21 against, 12 abstentions), while the Socialists (106 against), the Greens (44) and the Left (43) voted almost en bloc against. The nuance missing from many reports: without the three groups to the right of the EPP the text would still have passed, by 235 to 218.
Pressure from governments. In June, 19 governments signed a letter backing return hubs. Denmark, Austria, Germany, the Netherlands and Greece agreed a common model in Copenhagen on 4 September and aim for the first transfers in 2027, according to Euronews. Uganda and Rwanda are the candidates most often cited. The crossing of more than 70,000 people from Morocco into Ceuta at the end of July came after the Parliament vote, but it dominated the October Council.
The counter-hypothesis. An alternative reading says the vote is about political signalling, not efficiency, because orders to leave are already falling: by 12.8% year on year in the first quarter of 2026 and by 12% in the second, while returns are rising by 8–10%, according to Eurostat. Part of what the law promises was therefore happening without it. We weigh the counter-hypothesis as partly true. The trend exists, but the legal fragmentation was real, and its effect shows in the spread from 11% to 61% between states.
The consequences: four fronts
The legal timetable. Article 50 splits the text in two. Only a few articles apply from entry into force, among them the one on return hubs and those on minors. The rest, including extended detention, entry bans and the duty to cooperate, applies after 12 months, that is, in autumn 2027. The Commission has six months for the European Return Order form, and within three years the arrangements making the order available in the Schengen Information System must be in place. The law's first year is therefore mainly one of preparation.
Centres outside the EU. The only operational precedent, Italy's centres in Albania, has cost more than 670 million euros for about 500 people since March 2025, against the 3,000 a month that had been announced, according to the same Euronews analysis. Uganda has reportedly set conditions: Africans only, no criminal record and no political opponents, according to Agence Europe. Amnesty International warns that the centres risk becoming “black holes” for human rights. The fact that the law allows them does not mean a third country will host them at scale.
The courts. An appeal no longer automatically suspends removal, while the Charter of Fundamental Rights guarantees the right to an effective remedy. The tension between the two will almost certainly reach the EU Court of Justice through preliminary references from national courts, most likely once the main provisions start to apply. Spain could challenge the regulation directly, but as of 12 October we had found no announcement of an action for annulment.
Diplomacy with countries of origin. The regulation does not oblige Algeria or Morocco to take back their citizens. The lever remains visa, trade and development aid policy, which the regulation explicitly encourages. Here lies the risk Madrid flagged: a hard line towards Rabat, for instance, runs into cooperation at the Ceuta border. The return rate will genuinely rise only with the countries of origin that already cooperate.
What it means for Romania
Romania is a small player in these statistics. In 2025 it issued 2,860 orders to leave, against 3,895 in 2024 and 7,245 in 2023, and actually returned 775 people to a third country, against 2,700 in 2023, according to Eurostat. Romania ranks 22nd of the 31 reporting states. The falling figures point to fewer checks and fewer cases rather than greater efficiency: both the number of orders and the number of returns have fallen.
The profile of those concerned differs from the EU average. The top citizenships ordered to leave Romania in 2025 are Nepal (460), Sri Lanka (395), Moldova (295), Bangladesh (170), Türkiye (160) and India (145). Nepalis, Sri Lankans, Bangladeshis and Indians together account for 1,170 orders, or 41% of the total, and theirs is the profile of foreign workers brought in under the annual quota. The government approved a 2026 quota of 90,000 workers, 10,000 fewer than in 2025, according to Europa Liberă.
This is where the most likely practical consequence for Romania lies. A worker who loses their employer and is left without a right to stay will, from autumn 2027, fall under a common regime. An order issued in Bucharest will be visible in the Schengen Information System, and the entry ban can reach ten years. Because Romania has been a full Schengen member since January 2025, an order issued here could be enforced in Vienna or Berlin, and one issued there could be enforced here. Romanians in the diaspora are not affected: as EU citizens, they fall under free movement rules, not under this regime.
Romanian MEPs mostly voted in favour: 19 for, 4 against, 4 abstentions, and 6 did not vote, out of 33. Voting in favour were all nine EPP members present (PNL and UDMR), the six ECR members, Diana Iovanovici Șoșoacă, Luis Lazarus, Dan Barna (Renew) and Mihai Tudose, the only Socialist in favour. The PSD delegation split: Vasile Dîncu and Maria Grapini against, Gheorghe Cârciu, Gabriela Firea, Ștefan Mușoiu and Victor Negrescu abstaining, four absent. Vlad Vasile-Voiculescu (Renew) and Nicolae Ștefănuță (Greens) also voted against.
Testable predictions
| Prediction | Horizon | Probability | How it is checked | Verification source |
|---|---|---|---|---|
| The Return Regulation is published in the Official Journal of the EU | by 15.11.2026 | 85–95% | Regulation number and publication date | eur-lex.europa.eu |
| Denmark notifies its decision to implement the regulation in national law | by 1.04.2027 | 75–90% | Notification to the Council and the Commission | eur-lex.europa.eu, consilium.europa.eu |
| At least one member state signs an agreement for a return hub in a third country | by 31.03.2027 | 45–60% | Agreement or memorandum officially announced by a government | Statements by the Danish, Austrian, German, Dutch and Greek governments |
| First actual transfer of people to a return hub outside the EU under the new regulation | by 31.12.2027 | 30–45% | Official confirmation of the first transfer | Government statements, home-affairs.ec.europa.eu |
| Effective returns from the EU to third countries exceed 150,000 in 2026 | by 31.05.2027 | 50–65% | Annual value for the EU-27 | Eurostat, series migr_eirtn |
| Romania issues more orders to leave in 2026 than in 2025 (2,860) | by 31.05.2027 | 35–50% | Annual value for Romania | Eurostat, series migr_eiord |
| Spain brings an action for annulment of the regulation before the Court of Justice | by 31.01.2027 | 10–20% | Case registered with Spain as applicant | curia.europa.eu |
Method, definitions and figures
The votes. The plenary result, the breakdown by group and the vote of each Romanian MEP come from the European Parliament's roll-call record, as published by HowTheyVote.eu. The Council vote (23 in favour, one abstention, one against, two states not taking part) is as reported by Agence Europe. Romania's presence among the 23 follows by elimination.
The Eurostat figures. We use the annual series migr_eiord (orders to leave) and migr_eirtn (returns to third countries), in the versions updated on 30 and 25 September 2026 respectively. Eurostat's May release had given 491,950 orders and 135,460 returns for 2025. The difference comes from later revisions. The ratio of returns to orders is not a return rate in the strict sense: a person returned in one year may have received the order the year before, and unescorted voluntary returns are reported unevenly.
The legal text. The provisions cited (Articles 5, 8, 14, 28, 30 and 50, recitals 67–71) come from the adopted text, document PE-CONS 40/26 REV 1. The comparison with the 2008 directive covers the general rules, not every exception. Probabilities are editorial estimates in bands of at least ten percentage points.
Limits of this analysis
This analysis was written before the regulation's publication in the Official Journal, so the exact date of application is not yet fixed. We have no access to the five states' negotiations with Uganda and Rwanda, and the conditions attributed to Uganda come from press reports, not from a document. The figures for Romania do not show how many of those concerned were workers who came under the quota, and Eurostat does not publish that link.
The diagnosis would be refuted by a rapid rise in returns in the first year, before the main provisions apply, meaning a signalling effect stronger than the legal one, or by an EU-level framework agreement with Algeria or Morocco. The natural reassessment comes with publication in the Official Journal and with Eurostat's third-quarter data at the end of December.

.png?width=1200)
.jpg?width=1200)

