In an attempt to identify clinical predictors of future course and lithium response a more rigorously defined group of patients were studied than has previously been reported in the literature. Thus the study group comprised 98 bipolar 1 patients who had been hospitalised only once and satisfied DSM-III criteria for mania. Of the clinical factors studied, including the prescription of lithium, only the duration of illness predicted likelihood of re-admission. It was of little practical value, however, accounting for only 4% of the variance between groups. Although 49% of the patients were prescribed prophylactic lithium none of the variables studied predicted which patients relapsed despite adequate plasma levels of the drug. It would seem that prediction of outcome and response to lithium cannot be based on clinical factors.