Two recent studies [McMahon et al., 1995: Am J Hum Genet 56:1277-1286; Gershon et al., 1996: Am J Med Genet (Neuropsychiatr Genet) 67:202-207] reported an excess of maternal transmission in bipolar affective disorder in multiply affected families. In a sample of 130 families ascertained through a bipolar proband without regard to psychiatric family history we analysed the frequency of maternal (MAT) and paternal (PAT) transmissions, the morbid risk (MR) in relatives of transmitting mothers and fathers and the inheritance patterns in families with MAT vs. PAT transmission of the disease. In the total sample of 130 families we identified 39 families where the disease was transmitted from the paternal side (PAT families) and 35 families where the disease was transmitted from the maternal side (MAT families). Counting PAT and MAT transmissions in these unilineal families we found nearly equal numbers for both transmission types under a narrow (BP: bipolar disorder, schizoaffective-bipolar type disorder) and a broad definition (AFF: BP, recurrent unipolar depression) of the phenotype. The MRs for narrow and broad phenotypes were not significantly different in any type of PAT relative in PAT families vs. MAT relatives in MAT families. However, in PAT families there were two times more affected females than males with both disease models, while in MAT families there was no MR difference by relatives' sex. The transmission of BP was compatible with the Mendelian major gene model in PAT families and with the multifactorial model in MAT families. Extension of the relatives' phenotype led to borderline non-Mendelian major effects in PAT families and reproduced the multifactorial model in MAT families.