To address the observation of a secular trend in the incidence of major depression, we have evaluated prevalence of unipolar depression in first-degree relatives of unipolar depressed probands, all of whom were studied in the sleep laboratory. A threshold value of reduced (less than or equal to 65.0 min) or non-reduced (greater than 65.0 min) REM latency was used to define groups for both parents and siblings. Unipolar depression occurred at the same rate in both reduced REM latency siblings (57.1%) and parents (66.7%). Siblings with non-reduced REM latency had a higher rate of depression (36.8%) than non-reduced REM latency parents (0.0%). Implications for biological and environmental factors associated with liability for unipolar depression are discussed.