BACKGROUND Discrepancy between efficacy of prophylactic lithium and its effectiveness in ordinary clinical practice necessitates long-term follow-up data from specialised lithium clinics. Also, role of psychosocial factors in influencing the outcome is unclear. METHODS One hundred and eighteen patients of bipolar affective disorder attending a lithium clinic were followed-up for approximately 11 years (range 2-27 years). Demographic and clinical data, measures of social support and psychosocial stress were obtained at the intake in 1989-1990. Study design combined retrospective chart-review (till the time of intake) with prospective follow-up till July 1995. RESULTS On lithium, the patients had a mean of 0.43 relapses per year (manic, 0.26; depressive, 0.17) which was significantly less (p < 0.01) than the pre-lithium episode frequency. The figure for entirely relapse-free patients was 24%, and 62% had relapses up to one episode per year (median = 0.3 per year). Fifty-eight (49%) patients were good responders to lithium (relapses < or = 0.30 per year). In comparison to good responders, partial/poor responders had a significantly greater number of pre-lithium depressive episodes, poor lithium compliance, more psychosocial stress and lower social support at intake. These variables correlated well with relapses and explained 32% of the variance of the data. CONCLUSIONS Lithium had a definite prophylactic effect on long-term outcome. Social support and stressful life events are significant correlates of response to lithium. CONCLUSIONS Lithium prophylaxis of bipolar affective disorders seems justified though psychosocial factors appear to modulate its effectiveness. CONCLUSIONS Other psychotropic medications were used during relapse and the assessment of psychosocial factors was cross-sectional.