Sleep's relevance for long-term social functioning in psychiatric disorders has been widely overlooked so far. Here, we investigate social functioning in a transdiagnostic sample of 31 patients with severe mental illness, namely schizophrenia (n = 15) or major depression (n = 16), in relation to their polysomnographic sleep characteristics 6 (± 2.4) years earlier. In addition, cognitive performance at follow-up and clinical characteristics (i.e., severity of disorder-related symptoms and number of hospitalizations between baseline and follow-up) are assessed. Multiple regression analysis results in a model with slow-wave sleep (SWS) and number of hospitalizations as significant predictors accounting for 50% (R2 = 0.507; p <0.001) of the variance in social functioning. SWS remains a significant predictor of long-term social functioning throughout a series of refining analyses which also identify baseline functioning as an additional significant predictor, whereas diagnosis is non-significant. Also, the effect of SWS on social functioning is not mediated by number of hospitalizations as assessed by a bootstrapped mediation analysis. We thus conclude that duration of slow-wave sleep is a powerful predictor of long-term social outcome in psychiatric disorders. Also, we discuss the relevance of verbal memory, symptom severity, and diagnostic category for social functioning. Future studies should test this finding by using a prospective design, a bigger sample, optimized predictor variables, and a more diverse set of diagnoses. Moreover, it should be explored whether or not treating sleep disturbances in psychiatric illnesses independently improves long-term social functioning.