October 06, 2026
By Joanna Demopoulou
On 25 August 2026 the Grand Chamber held that Osman Kavala’s criminal conviction ‘should be regarded, under Convention law, as null and void’ (Kavala v. Türkiye (No. 2), § 309). Article 46 ECHR reasoning is ordinarily oriented towards obligations, results and measures: what the respondent State must achieve, and by what means, rather than towards what a national judgment is (Scozzari and Giunta v. Italy [GC], § 249; Kavala (No. 2), § 305). The sentence quoted above departs from that orientation, and it does not stand alone. The same paragraph requires Türkiye to eliminate the conviction’s consequences and points to a domestic reopening procedure whose premise is that the conviction still exists. This post asks what the nullity sentence adds to those two statements. The answer defended here consists of three parts. The sentence determines the conviction’s status under Convention law. No enforceable consequence can be identified that follows from it and not already from the duty to eliminate the conviction’s consequences. And it is not mere emphasis, because in execution it makes arguments that treat the conviction as a valid judgment with severable defects much harder to sustain.
Mr Kavala is a businessman and human rights defender, deprived of his liberty without interruption since 18 October 2017 (Kavala (No. 2), §§ 11-12). The accusations against him rested, as the Parliamentary Assembly of the Council of Europe put it in Resolution 2518 (2023), § 2, on activities falling within the ordinary role of a human rights defender.
Two principal streams of allegation ran against him: the Gezi Park protests of 2013, which he was said to have organised and financed, framed as an attempt to overthrow the Government by force (Article 312 of the Turkish Criminal Code), and the attempted coup of July 2016, framed as an attempt to overthrow the constitutional order under Article 309 Criminal Code (Kavala (No. 2), § 12). An espionage charge under Article 328 was added in March 2020 (Kavala (No. 2), § 13). The charges were brought in alternation, so that his detention never broke (Resolution 2518, § 1).
In Kavala v. Turkey (10 December 2019) the European Court found no reasonable suspicion that he had committed either offence and held that the accusations rested largely on conduct the Convention protects, such as meetings with journalists and foreign representatives, cooperation with Council of Europe bodies and campaigning against the sale of tear gas (Kavala v. Turkey, §§ 148 and 223). It found violations of Article 5 §§ 1 and 4 ECHR (Kavala v. Turkey, §§ 159 and 196) and, under Article 18 ECHR taken with Article 5 § 1, held that the detention had pursued an ulterior purpose: reducing him to silence (Kavala v. Turkey, § 232). Article 18 is engaged when a restriction the Convention permits, here detention for a prosecution, is applied predominantly for a purpose it does not allow; such findings are rare (Resolution 2518, § 4). Türkiye was to secure his immediate release (Kavala v. Turkey, § 240).
He was, however, not released. An İstanbul assize court acquitted him on the Gezi Park charge on 18 February 2020 and ordered his release; he was taken into custody the same day in the coup investigation and detained the next (Kavala (No. 2), §§ 30-31), whereafter the Council of Judges and Prosecutors opened an investigation into the three judges who had acquitted him (Resolution 2518, § 1). The acquittal was set aside on 22 January 2021 (Kavala (No. 2), § 38), and it is by that route, not the coup proceedings, that the Gezi Park allegations returned to trial.
On 2 February 2022 the Committee of Ministers referred to the Court the question whether Türkiye had failed to abide by the judgment of 10 December 2019. Under this infringement procedure (Article 46 § 4 ECHR) the Committee may, by a two-thirds majority, ask the Court to rule on non-compliance (Department for the Execution of Judgments, glossary). It is almost never used: the Kavala referral produced only the second such judgment (Resolution 2518, § 7). On 11 July 2022 the Grand Chamber held in Kavala v. Türkiye (infringement proceedings) that Türkiye had not complied. The Article 18 finding vitiated the measures resulting from the charges, and reclassifying the same facts could not alter that (Kavala (infringement proceedings), §§ 143 and 145). Nor had the State acted in good faith: it had kept him in detention for more than four years on facts the 2019 judgment had found insufficient, in a manner incompatible with the conclusions and spirit of that judgment (Kavala (infringement proceedings), §§ 171 and 173).
On 25 April 2022 the İstanbul 13th Assize Court convicted him under Article 312 and imposed aggravated life imprisonment; it treated the Article 309 coup charge as absorbed by that conviction (Kavala (No. 2), § 13) and acquitted him of espionage. His appeals were dismissed on 28 December 2022 and, by the Court of Cassation, on 28 September 2023, when conviction and sentence became final (Resolution 2518, § 1).
Two individual applications were still pending before the Turkish Constitutional Court when Strasbourg ruled: one lodged on 9 June 2022 against his pre-trial detention, one on 24 October 2023 against the detention resulting from the final conviction (Kavala (No. 2), §§ 70 and 72). They formed the basis of the Government’s non-exhaustion objection, and they are the subject of the dissent discussed below.
On 25 August 2026, by fifteen votes to two, the Grand Chamber found violations of Articles 3, 5 § 1, 6 § 1, 10, 11 and 18 ECHR. Under Articles 10 and 11, Kavala had been convicted without any finding that he committed, ordered or called for violence: violence by others was attributed to him through an assumed organising role, and protected activity became the material element of an offence carrying aggravated life imprisonment, so that Article 312 was extended unforeseeably and the interference was not ‘prescribed by law’ (Kavala (No. 2), §§ 171-172). The proceedings were unfair and the courts’ independence open to objectively justified doubt (§ 211); the pre-trial detention was arbitrary and the detention after conviction rested on a flagrant denial of justice (§§ 240 and 245); and the irreducible life sentence breached Article 3 (§§ 288-289). Under Article 18, the prosecution, the continued detention and the conviction were directed predominantly at punishing and silencing him (§ 279).
Article 46 § 1 ECHR binds States to abide by final judgments, under the supervision of the Committee of Ministers. The content of that obligation is settled: end the violation, make reparation for its consequences and restore as far as possible the situation before the breach, the choice of means resting in principle with the State (Scozzari and Giunta v. Italy [GC], § 249). In Kavala (No. 2), the Court restates this at § 305, citing Verein KlimaSeniorinnen Schweiz and Others v. Switzerland [GC], § 655, and Ilgar Mammadov v. Azerbaijan (infringement proceedings), § 150, adding at § 306 that in Türkiye execution is also a constitutional requirement, treaties on fundamental rights prevailing over conflicting statutes (Article 90 § 5 of the Constitution).
When the violation leaves no real choice, the Court may indicate the individual measure required, and does so here: at § 308, recalling the finding at § 245 that the detention following the conviction of 25 April 2022 rested on a flagrant denial of justice, it holds that release must be secured at the earliest possible date (Assanidze v. Georgia [GC], §§ 202-203; Del Río Prada v. Spain [GC], § 139). Everything there is expressed as something the State must achieve; nothing describes the legal quality of a national judgment.
Paragraph 309 then turns from release to the conviction itself, and it states three things. First, the conviction ‘should be regarded, under Convention law, as null and void’. Second, Turkish law provides a reopening mechanism ‘capable of erasing the effects’ of the conviction. Third, Türkiye must ‘eliminate the consequences’ of the conviction as soon as possible. The three statements sit in one paragraph, but they do different work.
The Court does not name the above-mentioned reopening mechanism. The relevant provision is Article 311 § 1 (f) of the Turkish Code of Criminal Procedure (Law no. 5271), which allows a final conviction to be re-examined after a violation found by this Court; it is identified here from the domestic law background and from the use now being made of it, and its operation was described in Mehmet Zeki Doğan v. Türkiye (No. 2), §§ 41-47 and 58, a case turning on the same provision. Whatever route is used must be capable of eliminating the conviction’s consequences, and reopening extinguishes nothing by itself: the conviction stands until the domestic court hearing the reopened case quashes or upholds it, and the request does not halt enforcement, although the court may suspend it (Article 312 of the same Code). The Court could not have ordered reopening in any event, since that lies outside its jurisdiction (Moreira Ferreira v. Portugal (No. 2) [GC], § 48; Guide on Article 46 (28 February 2026), § 18), and it neither quashes the conviction nor enters an acquittal. ‘Null and void’ is therefore not an annulment: nothing in the judgment removes the conviction from the Turkish legal order.
That is where the predicament lies. The Court has made a statement about what the conviction is, and has given that statement no operative expression and no stated consequence.
No enforceable consequence can be identified that follows from the nullity sentence alone. The result the Court requires in paragraph 309 rests on restitutio rather than on the status: the obligation restated at § 305 already calls for the elimination of the conviction’s consequences, § 309 recalls it before the nullity sentence appears, and the detention and the sentence are effects of violations already found. Three further features point the same way. There is no separate operative consequence: the individual measures (Kavala (No. 2), operative point 11), release and elimination of the conviction’s consequences, are what restitutio would require anyway, the general measures alongside them answer to the prevention of similar violations rather than to the nullity of this conviction, and the statement of status appears in none of those limbs. There is no doctrinal test: no criterion is given for when a conviction ceases to be one affected by Convention violations and becomes one lacking Convention law validity, and a category without a threshold is hard to apply. There is no authority: § 305 is supported by citation, § 309 by none, which proves nothing alone but does little to support the view that a new category is being introduced. Together these indicators show that the result expressly required of Türkiye would be the same if the sentence were deleted.
Nor does the sentence close off anything restitutio would otherwise leave open: no measure available under restitutio can be identified that § 309 shuts out. Thus, it does not create a new remedial category, a new obligation, or a nullity operating of its own force in Turkish law.
The rest of § 309, and the whole of § 305, allocate execution obligations: what the State must achieve, with the choice of means left to it. The nullity sentence is of another kind: it fixes the Convention law status of one judicial act without formulating a separate execution obligation or prescribing conduct. That status attaches to the conviction the Grand Chamber examined; the judgment does not address whether a reopened trial could affect it, or what Convention status a later domestic judgment would have. Its significance is interpretive: it can inform how those involved in execution understand that conviction, whether the Committee of Ministers or a domestic court asked to reopen the case, although how much weight it carries there the judgment does not settle.
In execution a State may argue about which effects are attributable to which violation, and about whether compensation is adequate redress for some. A statement that the conviction has no Convention law validity does not make those arguments legally unavailable, and nothing in the judgment says it does; it makes them substantially harder to maintain, since they presuppose a valid judicial act whose defects are to be weighed one by one, and the Court has declined to describe the conviction in those terms.
The İstanbul proceedings will be an early test of that difference. Following the judgment, Mr Kavala’s counsel applied to the İstanbul court that convicted him for suspension of enforcement, release and a retrial; on 9 September 2026 that court wrote to the Ministry of Justice asking whether the European Court of Human Rights had found violations in the case and requesting a Turkish translation (as reported on 11 September by Deutsche Welle’s Turkish service). A conviction characterised in Strasbourg as null and void is, in İstanbul, a live judgment awaiting the translation of the judgment that characterised it.
Two earlier cases test this reading. Ilgar Mammadov shows whether § 309 merely repeats an existing formula; Otegi Mondragón shows how wide the domestic margin is when the Court neither characterises the conviction nor prescribes a binding measure.
Ilgar Mammadov supports one proposition. Sitting under Article 46 § 4 ECHR, the Court held that the Article 18 finding in the first Ilgar Mammadov v. Azerbaijan judgment of 22 May 2014 vitiated any action resulting from the imposition of the charges (Ilgar Mammadov (infringement proceedings), § 189). The vitiating effect ran downstream, to the measures taken on the charges; the charges themselves were never annulled (Guide on Article 46, § 111), and the same formula returns in Kavala’s case (Kavala (No. 2), § 196). Paragraph 309 goes one step further in vocabulary, from the measures to the judicial act itself. That shift is why § 309 cannot be read as a paraphrase of Mammadov, and it is also the only difference the comparison establishes.
Otegi Mondragón and Others v. Spain supplies the contrast, and readers met it here in January, in Zelal Pelin Doğan’s analysis. The applicants, convicted of membership of the terrorist group ETA, had not been tried by an impartial tribunal; on 6 November 2018 the Court held the finding of a violation to be sufficient just satisfaction and indicated that a retrial would in principle be the most appropriate redress, without giving a binding indication as to execution (Otegi Mondragón and Others, §§ 74-75). The Spanish Supreme Court ordered a retrial; the Constitutional Court declared that order null and void as disproportionate once the sentences had been served; and on a subsequent application to Strasbourg a Committee of the Fifth Section declared the complaint inadmissible on 25 September 2025, holding that, absent a binding indication on execution, that reading was not a manifest error (Otegi Mondragón and Others (dec.), §§ 40-42). Where the Court says nothing about what the conviction is and prescribes no binding execution measure, the domestic authorities keep a wide margin over redress. In Kavala (No. 2) the Court does both of the things it withheld in Otegi: it characterises the conviction, and it prescribes the elimination of its consequences.
Judges Vehabović and Yüksel dissented on subsidiarity grounds, arguing that the application should have been rejected for non-exhaustion while the two Constitutional Court applications remained pending; in her separate opinion Judge Yüksel adds at § 11 that her disagreement under Articles 18 and 46 ECHR follows from that position. Neither engages with the characterisation itself.
When the Constitutional Court finds a violation arising from a criminal court’s decision, it sends the file back for a retrial so that the violation and its consequences may be removed (Law no. 6216, Article 50 § 2). Whether that remedy works is not theoretical: in two recent domestic cases, concerning Şerafettin Can Atalay and Tayfun Kahraman, criminal courts declined to give effect to Constitutional Court judgments (Kavala (No. 2), §§ 86-97), and a remedy that ends in a judgment the lower courts will not implement gives no redress.
Had the dissent prevailed there would be no § 309 to interpret, and the majority’s answer to it is confined. It reaffirmed that the individual application to the Constitutional Court is in principle effective, but held that its effectiveness was seriously compromised here, by years of detention, non-execution of the Court’s judgments, an acquittal that produced no result, and applications left undecided in what it calls procedural inertia (Kavala (No. 2), §§ 127-128). On these facts the majority is the more persuasive: subsidiarity presupposes a domestic remedy capable of producing a result, and applications pending since 2022 and 2023, before a Constitutional Court whose judgments in the Atalay and Kahraman cases the criminal courts had declined to implement, offered no such prospect.
The majority’s treatment of exhaustion is an exceptional, case-specific move rather than a doctrinal shift, and the same setting counsels against reading § 309 as a general rule: the characterisation arrives after an Article 18 finding covering the proceedings as a whole, a flagrant denial of justice, and years of unexecuted judgments, and which of those is necessary the judgment does not say.
Kavala (No. 2) claims no power to annul national judgments. For Türkiye the result is fixed: the consequences of the conviction must be eliminated and Kavala released at the earliest possible date. Paragraph 309 of the judgment adds a statement about the conviction, not a further obligation. That statement has practical weight, because an argument treating the conviction as a valid judgment whose defects can be weighed one by one is much harder to sustain against it. It does not, however, create a new remedial category, since no enforceable consequence can be identified that follows from it and not already from the duty to eliminate the conviction’s consequences. The rest will be decided elsewhere: by the Committee of Ministers in supervision, and by the İstanbul court now asked to reopen the case.
Strasbourg Observers is an academic blog that discusses recent developments at the European Court of Human Rights. The blog is run as an inter-university collaboration amongst the Dutch-language universities of Belgium. Associated research groups are the Human Rights Centre at Ghent University, the Centre for Government and Law at Hasselt University, the Leuven Centre for Public Law at KU Leuven, the Law and Development Research Group at the University of Antwerp and the Vakgroep Publiekrecht at the Vrije Universiteit Brussel.
Strasbourg Observers would not be possible without the valued work of countless guest bloggers. Visit the “Contact” page on our website to find out how to become one!
