Results of therapeutic trials with physostigmine in the treatment of Alzheimer disease (AD) have been inconsistent and controversy persists concerning safety and efficacy. In a double-blind, placebo-controlled, crossover study, patients received 6 weeks of oral physostigmine (OP) and placebo in random order. Twenty-nine patients with AD received as much as 16 mg/day of OP and were assessed with neuropsychological and functional measures. No significant cardiac side effects were noted, though other systemic adverse effects were noted, requiring dose reduction in four patients. There was a slight but significant improvement (12%) in performance on the selective reminding test with physostigmine and the memory performance was correlated with dosage. This improvement compares favorably with the 15% decrease in scores seen in an untreated comparison cohort followed for an equivalent time period. There was a trend toward an improvement in communication and a reduction in memory complaint. These results suggest that oral physostigmine is safe and may improve memory in AD.