Imagine your appeal lands before three judges. Two of them were once disciplined after a complaint by your husband, and one of those two presides. You ask them to step aside. They rule on the request themselves, and refuse.
That is what happened to Anna Poghosyan in Armenia. On 22 October 2026, in Anna Poghosyan v. Armenia, the European Court of Human Rights unanimously found a breach of her right to a fair hearing. The case turns on one word: impartial. Must a litigant prove that a judge is hostile, or only point to facts that would make a reasonable observer doubt them? The answer decides when the right to a fair hearing actually bites.
The story
Ms Poghosyan was sued over an unpaid loan, lost before a Yerevan district court in May 2016, and appealed.
In 2011, two of the three appeal judges, K.C. and T.N., had been disciplined (a reprimand and a warning) by Armenia's then Council of Justice after a complaint from her husband, a lawyer who had appeared before them. For K.C., who presided, it was a second sanction within a year. A third would have meant dismissal.
In January 2017 she asked both judges to withdraw. Each decided the challenge to their own impartiality, and both refused, in decisions that could not be appealed. Ten days later the court dismissed her appeal.
The definition
Article 6 of the European Convention on Human Rights guarantees a hearing by an "independent and impartial tribunal". The Strasbourg court reads "impartial" as having two limbs, set out by its 17-judge Grand Chamber in Micallef v. Malta and Morice v. France.
The subjective test asks whether a particular judge actually harboured personal prejudice. Judges are presumed impartial until proved otherwise, so this test is hard to satisfy. The objective test asks something different: whether, quite apart from the judge's behaviour, ascertainable facts give rise to a legitimate doubt. Here appearances count, because what is at stake is public confidence in the courts. Any judge about whom there is a legitimate reason to fear a lack of impartiality must withdraw.
The Court's own case-law guide on Article 6 sets out both as criteria for assessing impartiality. Neither is optional.
The shift and the debate
The Armenian judges' reading quietly collapsed impartiality into its first limb. Armenia's Judicial Code required recusal (a judge withdrawing from a case) where a judge "is biased", which, they insisted, could not be read as "could have been biased". They had shown no hostility, and had even granted Ms Poghosyan a deferral of court fees. The Code, they added, did not cover bias against a party's relatives, and her request, five months after she learned who was on the bench, looked like pressure. The Armenian Government defended that approach in Strasbourg, saying she had produced no evidence of objectively justified doubt.
That position has real force. Literal readings offer certainty, and the Code itself required judges to stay free of undue pressure. If a complaint by a litigant's spouse could disqualify a judge, complaints could become a way of choosing the bench.
The Court disagreed. An applicant, it said, need only show "an appearance of partiality supported by ascertainable facts", not prove actual bias. Disciplinary sanctions are a serious matter for a judge, and the close family link to the complainant would raise doubts for an objective observer. Because deliberations are secret, even one doubtful judge can taint a bench. Here there were two, a majority, including the president. The Bangalore Principles of Judicial Conduct, which the Court cited, point the same way: judges should step aside where a reasonable observer might doubt their impartiality.
Ironically, Armenia's Code opened by requiring recusal for objectively justified doubts, and listed its grounds only as examples. The judges treated one example as the whole rule. The Court also noted, without ruling on it, that Armenian law let challenged judges decide their own recusal with no appeal. It awarded Ms Poghosyan €3,600 for non-financial harm. The judgment is not yet final.
None of this means the judges were biased; Strasbourg found nothing to suggest personal prejudice. The point is definitional. If impartiality means only "not provably hostile", the safeguard works only when bias is visible, and the Court has long accepted that such evidence is hard to obtain. British readers will recognise the alternative: UK courts ask whether "the fair-minded and informed observer" would see a real possibility of bias, the test laid down in Porter v Magill. Because both tests demand facts rather than mere suspicion, they leave little room for tactical challenges. Strasbourg did not stretch the word "impartial". It refused to let it shrink.