The rapid-cycling variant of bipolar disorder constitutes about 15%-20% of all bipolar patients, and 72%-82% of these patients exhibit less than adequate response to lithium therapy. Valproate's spectrum of efficacy was examined in 78 patients with rapid-cycling bipolar disorder in a prospective, open, 15.8-month trial. Thirty patients received valproate monotherapy and 48 received combination therapy. Treatment assignment was nonrandomized and based on prior treatment history. A marked acute response was seen in 54% of the patients with mania, 87% of those with mixed states, and 19% of those with depression. Marked prophylactic responses were seen in 72% of manic patients, 94% of mixed states patients, and 33% of depressed patients. In addition, moderate acute antimanic responses were observed in another 31% of the patients, prophylactic antimanic responses in 17%, acute antimixed state responses in 0%, prophylactic antimixed state responses in 0%, acute antidepressant responses in 25%, and prophylactic antidepressant responses in mixed states in 34%. Pattern analysis was conducted to examine the spectrum of efficacy of valproate in various cells (e.g., the cohort of patients who had an acute antimanic response to the drug). Pattern analysis showed that 40% of the patients with a marked prophylactic antimanic response had a marked antidepressant response to valproate. However, among the patients with a marked antidepressant response to valproate, 91% had a marked antimanic response. The most common side effects of valproate in our study, as in earlier studies, were gastrointestinal problems (nausea, stomach cramps, diarrhea), tremors, lethargy, and hair thinning.(ABSTRACT TRUNCATED AT 250 WORDS)