Factors influencing the long-term outcome of affective disorders were investigated in 106 patients with an average length of course of 28 years. Univariate statistical analyses, stepwise multiple discriminance analyses and path analysis were applied. A direct unfavourable influence on the frequency of persisting alteration (evaluated according to the Global Assessment Scale) was found to have the low self-confident premorbid personality and a higher number of episodes during course, whereas only the variable "male" had such an influence on the social consequences of the illness. The variable "manic episode" (bipolarity) and "age at onset" had only an indirect influence, in that bipolar patients had more episodes, and younger patients more frequently had a bipolar course and therefore had more episodes. The unfavourable long-term outcome was due in those cases to the higher number of episodes.