Objectives: Mood stabilizers administered for bipolar disorder during pregnancy, such as valproic acid, can increase the risk of congenital anomalies in offspring. Valnoctamide is a valproic acid derivative associated with a decreased risk for congenital abnormalities in animals. The present study evaluated the efficacy and safety of valnoctamide monotherapy, compared to placebo, in the treatment of patients in an acute manic episode. Methods: A 3-week, double-blind, randomized, placebo- and risperidone-controlled, parallel group trial was conducted on 173 patients in an acute manic episode. Patients were randomized to receive valnoctamide 1500 mg/d (n=71), risperidone 6 mg/d (n=32), or matching placebo (n=70). The primary outcome measure was the change in Young Mania Rating Scale (YMRS) scores. Results: Valnoctamide did not differ significantly from placebo on any of the study endpoints (YMRS, Positive and Negative Syndrome Scale, and the Clinical Global Impression Scale for Bipolar Disorder [CGI-BP] scales; all P>.60). Mixed models for repeated measures showed that risperidone produced significantly more improvement than placebo in the overall bipolar disorder CGI-BP severity scale (P=.036), and the CGI-BP severity scale for mania (P=.021). The Kaplan-Meier survival curve revealed higher all-cause discontinuation rates (mainly due to lack of efficacy) in the valnoctamide group compared to the other study groups (P=.026). Patients with higher valnoctamide plasma levels had a numerically higher YMRS response, but this was not statistically significant. Conclusions: Valnoctamide was well tolerated at 1500 mg/d but lacked efficacy in the treatment of symptoms in patients with acute mania. Possible differences between the biological mechanisms of action of valproic acid and valnoctamide are discussed.