The intention of this study was to examine the possibility of predicting course of illness in manic patients, using a stress variable as a predictor, together with other more traditional indices. On the basis of information on three easily available parameters--family history, age at onset, and stress intensity--20 manic patients who had independent life events preceding an index admission were allocated into various risk groups and predictions were then made with regard to future course of illness for 5 years. On follow-up, the predictions were found to be correct in 85% of cases overall, and two high-risk factors were totally predictive.