A Procedure, Not a Presumption

Ceuta is a Spanish city of roughly 83,000 people on the North African coast, separated from Morocco by a land border and a stretch of sea. In July 2026 that border briefly stopped functioning as a filter. Spain's Interior Ministry reported that tens of thousands of people crossed into the enclave within a day or two — a figure early reporting placed between about 50,000 and 60,000, with a death toll that rose through the reporting window to around 67. Assessed — provisional These numbers are unsettled and diverge by source and by the window each source measures; we report the range, attribute it to the ministry, and treat no single figure as established.

What is not in dispute is the legal event the government put at the centre of its account. According to the Interior Ministry, the surge followed a ruling of Spain's Supreme Court on 29 June 2026, which barred authorities from summarily returning people arriving by sea at Ceuta and Melilla without an individualised process. Established The ministry added a causal claim of its own — that smuggling networks then exploited the ruling. That second statement is a government attribution, made by an interested party, and we flag it as a position rather than a fact. Assessed — government attribution

The important thing about that ruling is what it required. It did not declare the arrivals refugees. It did not presume the merits of anyone's claim. It required a procedure — that each person be dealt with individually before being sent back. That distinction, between a procedure and a presumption, is the whole of this piece.

1. What "refugee until proven otherwise" actually compresses

The slogan bundles at least three separate legal objects that operate independently and can be true or false separately: the prohibition on returning someone to persecution (non-refoulement), the rule governing who must prove what (the burden of proof), and the question of whether either applies at an irregular border crossing at all. Precision about which of these is engaged is not pedantry. It is the difference between an accurate criticism of the system and an inaccurate one.

Terminology carries the same load. A refugee is a person whose protection status has been recognised. An applicant or asylum seeker has a claim pending and no status yet. A person granted protection has cleared a decision in their favour. A person returned has been removed from the territory. Each word encodes a different stage and a different conclusion; letting one stand for another is how the slogan does its work. This piece keeps them apart.

2. Non-refoulement is procedural before it is substantive

The near-absolute core of the system is the prohibition on refoulement — returning a person to a country where they face persecution or serious harm. It is written into Article 33 of the 1951 Refugee Convention, Articles 18–19 of the EU Charter of Fundamental Rights, and — through Article 3 of the European Convention on Human Rights — into the case law of the Strasbourg court. Established

But read carefully, non-refoulement is procedural before it is substantive. What it forbids in the first instance is summary return — sending someone back without examining whether they would face harm. What it guarantees is that a claim gets looked at, and that the person is not returned while it is pending. It does not, by itself, guarantee any outcome. A claim can be examined in full compliance with non-refoulement and then rejected. The guarantee attaches to the process, not to the result.

This is exactly what the Spanish Supreme Court demanded on 29 June 2026, and it is why the ruling belongs to the procedural half of the doctrine, not the substantive one. It required Spain to examine before returning. It did not require Spain to admit, to recognise, or to grant.

3. The burden of proof is shared, not reversed

Here the slogan makes its sharpest error. "Refugee until proven otherwise" describes a reversed burden — the state must disprove the claim, and failing that, the person is treated as a refugee. European law does not work that way. Under Article 4 of the Qualification Directive (2011/95/EU), the applicant carries the duty to substantiate the claim — to present the facts, documents, and account on which it rests — while the determining authority assesses those elements in cooperation with the applicant. Established

That is a shared burden, not a reversed one. There is no legal presumption that an arrival is a refugee. A person who cannot substantiate a well-founded fear, or whose account does not meet the threshold, is refused — and, as Part 1 of this series documented from Eurostat data, a large share are: EU first-instance recognition often falls below half, and appeals overwhelmingly confirm rejections rather than overturn them, leaving a final-stage recognition rate near 21%. Established A system that rejects most appellants is not one that treats arrival as proof of status.

What is near-automatic on arrival is narrow and procedural: the right to lodge a claim, to have it examined, and not to be removed until it is decided. Mistaking that procedural entitlement for a presumption on the merits is the central analytical trap of the whole debate — and it is a trap because the honest version of the criticism is both narrower and more damning, as Section 6 sets out.

4. The limit at the border — N.D. and N.T. v. Spain

The procedural guarantee is not unlimited, and Spain is where its limit was drawn. In N.D. and N.T. v. Spain (Grand Chamber, 13 February 2020), the European Court of Human Rights considered two men who scaled the Melilla border fence in a large coordinated group and were immediately returned to Morocco. The Court held, by eleven votes to six, that there had been no violation of the prohibition on collective expulsion (Article 4 of Protocol No. 4), nor of the right to an effective remedy. Established

The decisive fact was the applicants' own conduct: they had bypassed the available legal entry points — the border post where a claim could be lodged — and instead forced the fence as a group. Having declined the lawful channel that existed, they could not, the majority reasoned, complain that their return without individual process was unlawful. Established

The ruling is narrow, and it is worth marking the narrowness, because it has been over-read in both directions. It does not license indiscriminate pushbacks; it turned on these applicants' avoidance of a genuine, accessible legal route. Contrast it with Hirsi Jamaa v. Italy (2012), where the same court extended non-refoulement to migrants intercepted at sea, who had no comparable lawful channel to use. The doctrine therefore already contains its own limiting principle: where an accessible legal entry point exists and is bypassed by irregular mass entry, immediate return can be lawful. The system is not the open door of the slogan.

5. The frame the crisis fell into — the 2024 Pact

Ceuta 2026 did not unfold under the old rules alone. The EU's 2024 Pact on Migration and Asylum — adopted by the Parliament on 10 April 2024 and the Council on 14 May 2024, in force since 11 June 2024 — had its main rules applied from 12 June 2026, six weeks before the July surge. Established The Pact introduces a border procedure: certain claims are processed at or near the external border within a compressed timetable — a twelve-week asylum phase, followed where relevant by a twelve-week return phase — while the applicant is kept in a controlled setting and treated, in law, as not yet having entered the territory. Established

The border procedure is still a procedure. It compresses the timetable and constrains movement; it does not abolish the examination or the non-refoulement floor beneath it. Whether it will, in practice, close the gap between rejection and return — the gap this series is really about — is the question the Pact's first real stress test at a place like Ceuta will answer, and it is a question this instalment leaves open.

6. What the law does — and does not — establish

Setting the pieces side by side, the documented legal reality is this:

• The law establishes a procedure. A right to lodge a claim, to have it examined, and not to be returned while it is pending — grounded in non-refoulement and the asylum acquis. Established
• The law does not establish a presumption of status. No reversed burden of proof; the applicant must substantiate, the authority assesses (Qualification Directive Art. 4). Most appeals confirm rejection. Established
• The procedural guarantee has a border limit. Irregular mass entry bypassing available legal channels can lawfully be met with return (N.D. and N.T. v. Spain). Established

Which locates the real, evidenced gap precisely. It is not that the system presumes people to be refugees — it rejects a large share of them. The gap is that rejection does not convert into return. As Part 1 documented, of the roughly half a million people ordered to leave EU territory in a recent year, only about one in five departed. Established That is a more damning finding than the slogan, and a more accurate one: the criticism that survives the evidence is about enforcement, not about a phantom presumption of the merits.

Bottom line: European law does not establish that an arrival is a refugee until proven otherwise. There is no reversed burden of proof: under the Qualification Directive the applicant must substantiate the claim and the authority assesses it — a shared burden — and most claims that reach appeal are rejected. What the law does establish is procedural: a right to have a claim examined and not to be returned while it is pending, subject to a real limit at the border, where bypassing available legal entry points can lawfully be met with return (N.D. and N.T. v. Spain). The Spanish Supreme Court's 29 June 2026 ruling — the government's stated trigger for the Ceuta surge — demanded exactly that procedure, not any presumption of status. The genuine, evidenced gap is not recognition but enforcement: rejection that does not convert into return. Whether that settlement is defensible is the merit-judgment this series has deferred.

Prediction — forward call

Part 3 of this series takes up the merit-judgment this instalment deferred: given that the law guarantees a procedure and removes only about one in five of those it orders to leave, is Europe's settlement between procedure and enforcement defensible on the evidence — and does the 2024 Pact's border procedure close the gap or merely relocate it?