THE 12-MONTH, double-blind, placebo-controlled trial of divalproex and lithium for patients with bipolar I disorder who are recovering from a manic episode, by Bowden et al for the Divalproex/Bipolar Study Group,1 was an extraordinary undertaking. It may be less important that it did not yield clear evidence for superiority of either drug over placebo than that it stimulates reconsideration of strategies for testing prophylactic therapy for bipolar disorder. Placebo controls have rarely been included in such studies since early testing for long-term effectiveness of lithium,2-4 and the divalproex study illustrates their continued importance. However, the challenge of enrolling 92 placebo-treated subjects no doubt had a major impact on the conduct of the divalproex study, and may well have determined its unexpectedly low recurrence rate with placebo (36/94 [38%])1 vs 12-month rates of 50% or higher in previous trials.3,4