Arrested in France on a warrant issued by a foreign government, you face a procedure that is fast, technical, and largely decided in a single hearing before the investigating chamber. Yet the outcome is never automatic. In November 2023, the criminal chamber of the Cour de cassation upheld a refusal to hand a man over to Ukraine, finding a real risk of a flagrant denial of justice in a country under martial law. In February 2026, the same chamber approved the surrender of another man to the same country, on the ground that Kyiv had this time given concrete guarantees. In October 2025, the Court quashed a decision to extradite an Austrian citizen to Brazil, both because he had not been brought to his own hearing and because France had failed to warn Austria first. These cases, all decided within the last three years, draw the map of how France really decides who gets put on the plane and who stays. This article explains that map: the human-rights filters that can block surrender, and the procedural weapons that can win the hearing.
I. Can France Refuse to Extradite You on Human-Rights Grounds?
French extradition law starts from a simple principle: where no international treaty says otherwise, the conditions, the procedure and the effects of extradition are set by the Code of Criminal Procedure, as Article 696 of that Code provides. Within that framework, the investigating chamber of the court of appeal gives a reasoned opinion, and the list of mandatory refusal grounds is closed and strict. Human-rights arguments therefore have to fit into precise legal boxes. Two of those boxes now generate most of the litigation: trial in a country at war, and exposure to death, a crushing sentence, or inhuman detention.
A. What if the country asking for you is at war? The Ukraine cases show how France weighs the risk
The key statutory box is Article 696-4 of the Code of Criminal Procedure, which states that extradition is not granted, and whose seventh paragraph covers exactly the wartime situation: “Lorsque la personne réclamée serait jugée dans l’Etat requérant par un tribunal n’assurant pas les garanties fondamentales de procédure et de protection des droits de la défense”. In other words, where the requesting state’s courts cannot ensure fundamental procedural guarantees and defence rights, France must say no. The fight is therefore about evidence: what would actually happen to this person, in this country, right now.
The leading illustration is the judgment of the criminal chamber of 7 November 2023, appeal no. 23-82.220 (ECLI:FR:CCASS:2023:CR01273). Ukraine sought the extradition of one of its nationals for aggravated misappropriation of property and aggravated money laundering going back to 2007 and 2013. The investigating chamber of the Chambéry court of appeal had issued an unfavourable opinion, and the public prosecutor appealed. The Cour de cassation dismissed the appeal and let the refusal stand. It endorsed findings that Ukraine had exercised its right to derogate from its treaty obligations, particularly under Articles 5, 6, 8 and 13 of the European Convention on Human Rights, and that the wartime amendment of Article 615 of the Ukrainian Code of Criminal Procedure allowed detention ordered by an authorised officer rather than an investigating judge, as well as suspension of investigations until the end of martial law, with the prosecutor deciding on continued pre-trial detention. Those restrictions, the judges found, applied across the whole of Ukrainian territory, not only in occupied areas, and with courts suspending non-urgent cases, a trial within a reasonable time was not guaranteed. The decisive holding was that serious and proven reasons to believe supported a real risk of a flagrant denial of justice for that person, given the martial-law pre-trial detention rules then in force and an armed conflict deeply disorganising the Ukrainian justice system and stalling proceedings. That is the test, and it is assessed concretely, person by person, on the file as supplemented by additional information ordered by the chamber. In that case the chamber had ordered such further inquiries in January 2023 before ruling, which shows how the defence can force the court to look beyond paper guarantees and test, with evidence, what surrender would mean in practice for the individual concerned.
That refusal was not a blanket ban on surrenders to Ukraine. On 29 October 2025, the Ninth Investigating Chamber of the Paris court of appeal again refused surrender in a Ukrainian case, on similar reasoning: martial law covering the entire territory, derogations from the Convention, wartime criminal procedure departing from ordinary safeguards, and diplomatic assurances considered too general to compensate. Yet barely three months later, the Cour de cassation confirmed the opposite outcome. By a judgment of 10 February 2026, appeal no. 25-86.478 (ECLI:FR:CCASS:2026:CR00172), it dismissed the appeal of a Ukrainian national whose surrender had been approved by the Paris investigating chamber on 17 September 2025. He was wanted under a 2023 arrest warrant for large-scale cannabis cultivation and drug offences committed in 2020, carrying up to ten years of imprisonment. The defence had invoked the same martial-law derogations from Convention rights. This time, however, the chamber had obtained detailed replies from Kyiv: assurances on detention conditions and access to a lawyer, a guarantee of “un examen judiciaire équitable et public dans un délai raisonnable par un tribunal indépendant et impartial, ainsi que le droit à la défense et à l’assistance d’un avocat”, and confirmation that only an investigating judge would decide on pre-trial detention. On that basis, the Cour de cassation held that the chamber “a souverainement écarté l’existence du risque d’atteintes aux droits fondamentaux excipé par la personne réclamée.” Same requesting state, same war, opposite result: the difference was made entirely by the quality of the concrete guarantees in the file.
The French position sits inside a wider European split that defence lawyers now exploit openly. Italian courts, after initially accepting surrenders based on assurances tied to calm regions, increasingly treat the country-wide and unpredictable effects of the war as an established fact and regard general Ukrainian assurances as insufficient. The Rotterdam District Court refused surrender twice, in October 2025 and March 2026, on detention conditions, air-raid exposure and lack of an effective remedy under Articles 2, 3 and 13 of the Convention. London, by contrast, broke ranks: in a March 2026 judgment published by the judiciary (Ukraine v Dmytruk), the Westminster Magistrates’ Court refused extradition on private-life grounds but accepted Kyiv’s assurances on fair trial, reasoning that martial-law provisions do not bite where ordinary law keeps operating. For anyone arrested in France on a Ukrainian warrant, the practical lesson is therefore clear: the case is won or lost on evidence about detention, judges and delay after surrender, and the defence must attack vague assurances line by line while the prosecutor tries to top them up through supplementary information.
B. What if you face the death penalty, a life sentence, or inhuman detention after surrender?
The second great filter is Article 3 of the European Convention on Human Rights, which prohibits torture and inhuman or degrading treatment or punishment. France cannot knowingly send someone toward such treatment. The Cour de cassation recalled the architecture of this control in a Russian case decided on 13 December 2022, appeal no. 22-80.610 (ECLI:FR:CCASS:2022:CR01558): a state may incur responsibility where there are serious and proven reasons to believe the person faces a real risk of the death penalty, torture, or inhuman or degrading treatment, and the same logic applies to a risk of a flagrant denial of justice under Article 6. Russia, the Court added, had ceased to be a party to the Convention on 16 September 2022, so that the Convention-based assurances Moscow had attached to its request had lapsed. Yet the appeal was dismissed: the person had only invoked risks to private and family life and religious freedom under Articles 8 and 9, rights which, outside the parties to the Convention, cannot by themselves block surrender. Two lessons follow. First, assurances die when the legal order that backed them dies, and any guarantee drafted before September 2022 and pointing at the Convention is worthless against Russia today. Second, only risks of the highest order, under Articles 2, 3 and 6, stop a surrender; lesser grievances do not.
Where the death penalty is on the table, the control is at its strictest. On 30 January 2024, appeal no. 23-83.549 (ECLI:FR:CCASS:2024:CR00067), the Court quashed a favourable opinion given by the Aix-en-Provence chamber for surrender to Tunisia of a man prosecuted for premeditated murder. The chamber had relied on Tunisian assurances plus the country’s long de facto moratorium, noting that nobody had been executed in more than thirty-two years and that Tunisia kept voting for the United Nations moratorium resolutions. The Cour de cassation swept that reasoning aside: “il appartient à la chambre de l’instruction de s’assurer que l’Etat requérant prend l’engagement dénué de toute ambiguïté que cette peine ne sera pas appliquée à la personne réclamée.” An unambiguous commitment, covering this specific person, is required; general considerations about a moratorium, which the chamber itself had described as symbolic, cannot satisfy the test. Anyone facing a capital charge must therefore demand a personal, written, unconditional non-application guarantee, and attack anything less as legally weightless. Readers facing this specific scenario will find a fuller treatment of the punishment test in our analysis of how France tests the punishment before extraditing.
Even short of death, a sentence can be so heavy that surrender becomes inhuman. On 6 May 2025, appeal no. 24-85.773 (ECLI:FR:CCASS:2025:CR00551), the Court confirmed a surrender to South Korea of a man facing life imprisonment for large-scale financial and stock-market fraud worth tens of billions of won. The rule it stated is now the reference: “l’extradition doit être refusée lorsque la personne concernée démontre qu’elle est exposée à un risque réel de voir prononcer contre elle une peine manifestement disproportionnée dans l’Etat requérant, cette notion devant faire l’objet d’une interprétation stricte et ne pouvant être retenue que dans des cas très exceptionnels”. The mere fact that the foreign sentence would be heavier than the French one proves nothing; the chamber assesses disproportionality against the nature and gravity of the facts and the conditions under which the foreign court will sentence, and “La Cour de cassation ne peut substituer son appréciation à celle des juges.” In practice, this means the disproportionality argument must be built fact by fact at the hearing, with comparables and sentencing practice, because nothing can be repaired afterwards by the supreme court. Life sentences for economic crime, calibrated by statute to the amounts at stake, survived the test; genuinely aberrant sentences remain the rare cases where it bites.
II. How Do You Fight an Extradition Request Before the French Courts?
Substance is only half the battle. The French procedure concentrates everything into one hearing, and the defence that masters its mechanics often achieves what human-rights rhetoric alone cannot. When the requested person tells the public prosecutor that he does not consent to surrender, the investigating chamber is seized without delay, and Article 696-15 of the Code of Criminal Procedure sets a brutal timetable: “La personne réclamée comparaît devant elle dans un délai de dix jours ouvrables à compter de la date de sa présentation au procureur général.” The chamber then gives a reasoned opinion within one month of the appearance, unless it orders further information. That opinion must be unfavourable where the legal conditions are not met: “Cet avis est défavorable si la cour estime que les conditions légales ne sont pas remplies ou qu’il y a une erreur évidente.” And review is narrow: “Le pourvoi formé contre un avis de la chambre de l’instruction ne peut être fondé que sur des vices de forme de nature à priver cet avis des conditions essentielles de son existence légale.” Only formal defects capable of depriving the opinion of the essential conditions of its legal existence can support an appeal to the Cour de cassation. Every procedural flaw must therefore be identified, stated on the record, and pushed at the hearing itself.
A. Why your presence at the hearing can decide the case before it starts
The most spectacular recent illustration came on 15 October 2025, appeal no. 24-86.904 (ECLI:FR:CCASS:2025:CR01322). Brazil sought a man for firearms offences; the Fort-de-France investigating chamber gave a favourable opinion on 17 September 2024 without him, because a prison guards’ protest had made it impossible to bring him to court, and the chamber, invoking imperative hearing deadlines, had decided to proceed at the end of the sitting. The Cour de cassation quashed the decision. It recalled first what the hearing is for: “tout accusé a le droit de se défendre lui-même ou d’avoir l’assistance d’un défenseur de son choix et, s’il n’a pas les moyens de rémunérer un défenseur, de pouvoir être assisté gratuitement par un avocat d’office, lorsque les intérêts de la justice l’exigent”, the words of Article 6, paragraph 3 of the Convention. It then held: “En émettant un avis favorable à l’extradition en l’absence du demandeur, qui n’avait pu être extrait, en raison d’une circonstance qui n’était pas extérieure au service de la justice, et alors que sa comparution pouvait être reportée, jusqu’au 23 septembre à minuit, la chambre de l’instruction n’a pas justifié sa décision.” A strike inside the prison system is not an event external to the justice system, and where the appearance could still be postponed within the time limit, judging an absent person is a fatal defect. The message to every defence team is blunt: never accept that the chamber proceeds without your client, demand the postponement on the record, and make the court explain in writing why it refuses.
The same judgment pattern shows a second procedural lever: the request for supplementary information. In the Ukrainian cases, the chambers repeatedly paused to demand answers on detention conditions, access to counsel and trial delays, and in the Korean case the chamber waited for Seoul’s replies before ruling. Each order buys time and forces the requesting state to commit itself in writing, which creates the very material on which assurances are later tested. Defence memoranda should therefore never simply deny the risk; they should list precise questions, the chamber should be asked to put them formally, and vague replies should be exhibited as proof that no concrete guarantee exists. It was the absence of concrete answers that killed the Chambéry surrender in 2023, and the presence of detailed ones that saved the Paris surrender in 2026.
B. Why your European passport forces France to call your home country first
The October 2025 Brazilian case quashed the Fort-de-France decision on a second, independent ground that every European Union citizen arrested in France should know. The requested man was an Austrian national, and the chamber had approved his surrender to Brazil without checking whether Austria had ever been warned. The Cour de cassation recalled the rule flowing from the Court of Justice’s Petruhhin judgment of 6 September 2016 and its By judgment of 17 December 2020: a Member State receiving a third-country extradition request for another Member State’s citizen must inform that citizen’s home Member State and, where the home state so requests, hand the person over to it under the European Arrest Warrant Framework Decision of 13 June 2002. Loyal cooperation requires the requested state to pass on not just the existence of the request but all the detailed legal and factual material supplied by the third country, to keep the home state informed of any relevant change in the proceedings, and to let it decide whether to prosecute and issue a European arrest warrant itself. Because none of that had been done for Austria, the favourable opinion was quashed in its entirety and the case sent back to differently composed judges. Readers concerned by post-Brexit surrenders between France and the United Kingdom will find the mirror-image mechanics explained in our study of British warrants executed in France since Brexit.
This European shield sits next to a purely French one that is even stronger: nationality. The very first paragraph of Article 696-4 provides that extradition is not granted “Lorsque la personne réclamée a la nationalité française, cette dernière étant appréciée à l’époque de l’infraction pour laquelle l’extradition est requise”. A French citizen cannot be extradited, full stop, with nationality assessed at the time of the offence. For Union citizens who are not French, the Petruhhin route is the functional substitute: it does not forbid surrender outright, but it forces France to offer the home country the first move, and any file in which that step was skipped is vulnerable on appeal. Counsel should therefore verify, from the first meeting, the client’s nationality history, demand proof that the home Member State was informed with the complete file, and plead the omission expressly, since the Cour de cassation treats it as a self-standing ground for quashing.
Conclusion
France extradites, but it does not rubber-stamp. The cases of 2022 to 2026 trace a consistent method: the investigating chamber must look at what concretely awaits the person, order additional information where the file is thin, and refuse surrender where the guarantees do not hold up, whether the requesting country is at war, retains the death penalty, or has left the Convention system. The Cour de cassation polices that method from above, quashing opinions given in the person’s absence, opinions that skip the warning to the home Member State, and opinions that dress general context up as personal guarantees, while letting stand the surrenders where concrete, individualised assurances answer each documented risk. For the person arrested, the strategy follows directly: contest early, demand written answers to precise questions, appear and be heard, invoke European citizenship where it exists, and force every assurance to name its beneficiary. Surrender is a test of evidence, and in France the defence gets to set much of the paper on which it is marked.