First and second degree relatives of 99 probands with affective disorders (49 unipolar and 50 bipolar subjects) were studied. The high risk values obtained for affective disorders were shown to be compatible with those found by other authors, although the prevalence of the illness in the population of Lombardy appears to be much lower than in other countries. Very low rates of suicide and alcoholism were found in our sample. Data obtained by analysis of the affected pairs of relatives rule out the hypothesis of a dominant X-linked gene if the bipolar and the unipolar forms are considered genetically separated entities. Results compatible with a polygenic condition, partially shared by bipolar patients, were found using Slater's and Smith & Falconer's methods. Our data, however, cannot rule out the dominant hypothesis.