Sentencing disparities and the EAW in Tagu (C-583/24)


Three grams of cannabis and four MDMA tablets. For bringing them into Romania with his wife, DZ received a final sentence of seven years. He was arrested in the Netherlands, where that same conduct could attract just one to three weeks in prison. The Amsterdam District Court asked the CJEU whether the seven-year sentence was disproportionate to the offence and, if so, whether it could refuse surrender on that ground.
In Tagu (C-583/24, 3 November 2026), the Grand Chamber of the Court of Justice held that a breach of the right to a proportionate penalty under Article 49(3) of the EU Charter can, exceptionally, justify refusing a European arrest warrant. However, it also concluded that, in the specific circumstances of the case, the Dutch court could not refuse surrender on that ground.
Why Amsterdam questioned the seven-year minimum
During the EAW proceedings, DZ told the Amsterdam District Court that the drugs were for his wife, who used them to relieve pain. Whatever their precise purpose, the Dutch judges were satisfied that neither spouse had intended to traffic them. Yet under Romanian law at the time, importing a high-risk drug such as MDMA carried a prison term of seven to fifteen years. DZ had received a seven-year imprisonment sentence.
The Dutch court accepted that none of the usual grounds for refusing surrender applied in that case. What troubled it, in terms of proportionality, was that, once the Romanian judges had ruled out mitigation, they could go no lower than seven years, regardless of how little the couple had brought across the border and whatever their reason for doing so.
The Court of Justice had already dealt with differences in sentencing in Procureur général près la cour d’appel d’Angers (C-168/21) of 14 July 2022, discussed in more detail in an earlier eXtradando post. One of the sentences Italy sought to enforce was a ten-year prison term for “devastation and looting” during the Genoa G8 protests. The conviction rested on seven acts, two of which, in the French court’s view, would not have been offences in France. The CJEU held that, once double criminality was established, the French judges could not rely on that discrepancy to reassess the proportionality of the ten-year term.
The Amsterdam court approached the issue from a partially different angle. Even if all the ordinary conditions for surrender were met, could Article 1(3) of Framework Decision 2002/584, which preserves the duty to respect fundamental rights, justify refusing surrender if the sentence DZ would have to serve was disproportionate to his offence under Article 49(3) of the Charter?
What the executing judge can examine
The CJEU applied the two-stage inquiry developed in Aranyosi and Căldăraru for prison condition.
Applied to sentencing, the first question is whether objective, reliable, specific and up-to-date evidence establishes a systemic or generalised deficiency in the issuing State’s law or practice, or a deficiency affecting an objectively identifiable group to which the requested person belongs. Only then may the executing judge ask whether there are substantial grounds for believing that, because of that deficiency, the person would face a real risk of serving a disproportionate sentence if surrendered. Even though DZ’s seven-year sentence was already final, the Dutch judges still had to establish that wider deficiency before they could examine his individual case
In this respect, the CJEU distinguished the case from E.D.L. (C-699/21), where the risk that surrender posed to a seriously ill person arose from his own state of health and could be assessed directly by the executing judge (Tagu, §§ 61-71).
The CJEU, thus, adapted the first stage of Aranyosi to sentencing by setting two cumulative conditions: (i) the law or judicial practice under which the sentence was imposed must give rise to a manifest and serious breach of proportionality, and (ii) there must be no effective judicial remedy in the issuing State capable of correcting that breach.
The Court held that whether a penalty crosses that threshold is a question of EU law. Framework Decision 2004/757 on drug trafficking leaves each Member State free to decide whether and how to punish someone who imports drugs exclusively for their own use and, if so, how severely. Therefore, the disparity between the weeks cited by the Dutch prosecutor and the seven years imposed on DZ could not, therefore, establish alone neither that Romanian law or sentencing practice entailed a manifest and serious breach of Article 49(3) of the Charter, nor that Romania lacked an effective judicial remedy to correct such a breach (Tagu, §§ 72-77 and 82).
The judgment tried to explain what may make a breach sufficiently serious. If the law allows a prison term so out of proportion to the offence that it amounts to inhuman or degrading punishment under Article 4 of the Charter, the breach is plainly serious enough (Tagu, § 83). Likewise, if the law takes no account of the gravity of the offence when setting the penalty and the judge has no room to choose a shorter term within the statutory range, go below the minimum or reclassify the offence, that too amounts to a manifest and serious breach of Article 49(3).
A comparison helps here. In G. ST. T. (C-655/21), judgment of 19 October 2023, a Bulgarian court asked whether its own law could prescribe a minimum of five years’ imprisonment for trademark infringement where the conduct was repeated or caused significant harmful effects. The CJEU held that an infringement could be repeated and still have very little commercial impact. Yet the sentencing judge could go below five years only if the mitigating circumstances were exceptional or numerous. Given the range of conduct the offence covered, that narrow exception left judges too little room to adapt the sentence to what the defendant had actually done. The Court therefore held that Article 49(3) precluded the five-year minimum (§§ 80-86).
In DZ’s case, Luxembourg read the seven-year term at the bottom of the range as an indication that the Romanian judges appeared to have taken the quantity and purpose into account. Romanian law also allowed a reduction below the minimum for mitigating circumstances and a different legal classification where the evidence supported it. In any case, according to the CJEU, whether the facts justified either course was for the Romanian courts to decide.
Accordingly, DZ’s proportionality objection therefore failed at the first stage of the test.
Where Strasbourg draws the line
In Harkins and Edwards v. the United Kingdom [GC], judgment of 17 January 2012, two men facing extradition to the United States on murder charges argued that the sentences they faced would breach Article 3 ECHR. Strasbourg accepted that, in principle, a grossly disproportionate sentence can amount to prohibited punishment under Article 3 ECHR. However, in that case, the Court held that neither applicant risked receiving such a sentence (ivi, §§ 133-134 and 139-142). The European Court thus looked directly at what each man faced in the US, unlike Tagu where the CJEU held that domestic courts may examine an RP’s individual position only after establishing a wider deficiency in the issuing MS.
The closest comparison on the evidence is Bijan Balahan v. Sweden, judgment of 29 June 2023, cited by Advocate General Richard de la Tour in his Opinion in Tagu (§ 77). The applicant argued that, given his previous convictions, California’s Three Strikes Law could leave him serving sixty-one years before he became eligible for parole. But that calculation depended on his being convicted on every charge and on the sentencing enhancements being applied in full. Additionally, both the prosecutor and the judge had some discretion over those enhancements. For his part, the applicant was unable to point to anyone with a similar record who had received such a term for comparable conduct. On that evidence, The Strasbourg Court found no real risk that he would receive a grossly disproportionate sentence (§§ 60-68).
The drug sentences in Willcox and Hurford v. the United Kingdom, decision of 8 January 2013, offer a further comparison with DZ. Two British prisoners asked to serve the remainder of their Thai sentences in the United Kingdom. Once back in the UK, they challenged their continued detention, pointing to the much shorter terms they would have received in England. There, the Court held that “a grossly disproportionate sentence could amount to ill-treatment contrary to Article 3. However, gross disproportionality is a strict test and it will only be on rare and unique occasions that the test will be met”. In that case, the Strasbourg Court observed that both sentences fell within the English statutory maxima and that both men had requested their transfers. It expressly distinguished voluntary transfer, which improved their position, from extradition, where the requested State can prevent exposure to the punishment. It thus held that the Article 3 complaints were manifestly ill-founded (§§ 74-81).
How the Italian Court of Cassation has approached the same problem
The Italian Court of Cassation has long recognized that a foreign sentence can bar extradition if it is wholly unreasonable and manifestly disproportionate to the offence. In practice, challenges on that ground have rarely succeeded.
In judgment no. 38137/2008 of 24 September 2008, Romania sought to enforce a three-year sentence for entering residential premises at night and stealing maize cobs. The Supreme Court found no manifest disproportion in three years for aggravated theft and unlawful entry. In a later case, judgment no. 5400/2009 of 23 January 2009, a young man with no previous convictions challenged a three-year Romanian sentence for taking a bicycle he valued at €50. The Supreme Court again upheld surrender on grounds that the offence was a robbery and that the price of the bicycle said little about the violence involved in taking it.
On the other hand, the Cassation refused extradition in judgment no. 21988/2013 of 15 April 2013, another Romanian case. The RP had received three and a half years for a series of chicken thefts from private courtyards. By the time his case reached the Italian courts, he was supporting two daughters in Italy, one aged nine and the other a newborn. His wife’s seasonal earnings would not cover the family’s basic needs. Sixteen years after the thefts, the Court held that the children’s need for their father’s support outweighed Romania’s limited interest in enforcing the sentence. It thus quashed the decision authorizing extraditionwithout remittal.
In judgment no. 2037/2019 of 5 December 2018, published on 16 January 2019, China was seeking extradition for trial on charges of fraud. Under Chinese law, a particularly large sum could attract more than ten years’ imprisonment or life. The Supreme Court remitted the case back before the lower court because no proportionality assessment had been carried out as to what would guide the choice between a term of years and life.
More recently, the issue has come up in two US extradition case. In the first case, the RP was sought on charges of conspiracy to launder money carried a twenty-year maximum, against seven years for the Italian offence of criminal association. The US judge could impose less, and the sentence was open to review under the Eighth Amendment. Applying Article 49 of the Charter by virtue of the EU-US extradition agreement, the Cassation rejected the sentencing objection (judgment no. 24061 of 13 May 2026). In the second case, aggravated identity theft carried a mandatory two years on top of any term for the underlying fraud. The Court upheld extradition after considering the sentence as a whole, including the discretion the judge retained on the fraud counts (judgment no. 25583 of 8 July 2026).
Conclusion
Tagu leaves open whether seven years was too much for three grams of cannabis and four MDMA tablets. Before the District could examine DZ's sentence, it needed evidence that Romanian law or practice exposed others in his position to manifestly excessive terms and left them without an effective judicial remedy. For Italian courts too, the practical significance of extending the Aranyosi and Căldăraru test to sentencing lies in that first step. Most objections are likely to fail there, before the executing judge ever considers whether the punishment fitted the offence.



Comments