Tyrosine hydroxylase (TH), the rate-limiting enzyme in the metabolism of catecholamines, is considered a candidate gene in bipolar affective disorder (BPAD) and has been the subject of numerous linkage and association studies. Taken together, most results do not support a major gene effect for the TH gene in BPAD. Genetic and phenotypic heterogeneity may partially explain the difficulty of confirming the exact role of this gene using both association and linkage methods. Four hundred one BPAD patients and 401 unrelated matched controls were recruited within a European collaborative project (BIOMED1 project in the area of brain research, European Community grant number CT 92-1217, project leader: J. Mendlewicz) involving 14 centers for a case-control association study with a tetranucleotide polymorphism in the TH gene. Patients and controls were carefully matched for geographical origin. Phenotypic heterogeneity was considered and subgroup analyses were performed with relevant variables: age at onset, family history, and diagnostic stability. No association was observed in the total sample or for subgroups according to age at onset (n = 172), family history alone (n = 159), or high degree of diagnostic stability and a positive family history (n = 131). The results of this association study do not confirm the possible implication of TH polymorphism in the susceptibility to BPAD.