Mood disorders are known to cluster within families, but the mode of transmission remains largely unknown. The purpose of our analysis was to determine whether selection of a sample that was homogeneous in its response to an antidepressant provided stronger evidence for a single major locus. Complex segregation analysis was applied to a sample of 171 Italian families of bipolar and unipolar probands that were responsive to the antidepressant fluvoxamine. We used regressive logistic analyses to determine the best fit from among environmental, arbitrary Mendelian, dominant, recessive and additive models. For the 171 affective families with probands that were responsive to the antidepressant fluvoxamine, a Mendelian model of inheritance was rejected. When considering 68 families of bipolar probands, the best fit was obtained for a Mendelian dominant model of transmission. The identification of a Mendelian mode of transmission in bipolar subjects who were selected according to their response to fluvoxamine supports the use of a pharmacological criterion as a tool for identifying true genetic disorders.