We treated 149 patients meeting criteria for age‐associated memory impairment (AAMI) for 12 weeks with a formulation of phosphatidylserine (100 mg BC‐PS tid) or placebo. Patients treated with the drug improved relative to those treated with placebo on performance tests related to learning and memory tasks of daily life. Analysis of clinical subgroups suggested that persons within the sample who performed at a relatively low level prior to treatment were most likely to respond to BC‐PS. Within this subgroup, there was improvement on both computerized and standard neuropsychological performance tests, and also on clinical global ratings of improvement. The results suggest that the compound may be a promising candidate for treating memory loss in later life. NEUROLOGY 1991;41:644‐649