Using a longitudinal life-table analysis, we assessed the efficacy of lithium alone, administered within the context of a naturalistic clinical setting, by calculating the probability of patients remaining free of an affective episode (manic or depressive) over a five-year course. In addition, for those who suffered a manic or depressive relapse, we attempted to analyse the subsequent course of patients who suffered a manic/hypomanic or depressive relapse and were then restabilised on lithium plus either a neuroleptic, carbamazepine, or a benzodiazepine, or lithium plus an antidepressant. Lithium alone offered an average 83% probability against an affective relapse after one year, 52% after three years, and 37% after five years. For patients who failed on lithium alone, it appeared that combination treatment offered greater protection against subsequent affective relapse than the initial course on lithium alone.