Objective: A multicenter trial to evaluate the efficacy of controlled- release physostigmine salicylate, a cholinesterase inhibitor, was conducted in 1,111 mild-to-moderate Alzheimer's disease (AD) subjects. Design: During dose titration, subjects received 18, 24, or 30 mg of physostigmine or placebo daily. After a 2-week washout period, 366 subjects with putative improvement were randomized to receive either placebo or their best dose of physostigmine in a 6-week double-blind trial. Nonresponding patients (439) were randomized to receive in a separate double-blind trial either placebo or their highest tolerated dose of physostigmine. The primary efficacy measures included the cognitive subscale of the Alzheimer Disease Assessment Scale (ADAS) and a Clinical Global Impression of Change (CGIC). Secondary measures included the Mini-Mental State Examination and two activities-of-daily- living scales. Results: At the end of the 6-week double-blind phase, physostigmine-treated patients scored 1.75 points higher than placebo- treated patients on the ADAS (p = 0.003) and 0.26 points higher on the CGIC (p = 0.012) in the intent-to-treat analysis. There was no significant improvement on the secondary outcome measures. Patients failing to respond to physostigmine during the dose titration phase failed to respond on any of the outcome measures during the double-blind period of re-exposure. Common adverse events included nausea, vomiting, diarrhea, and anorexia. There were no significant changes in liver function tests. Conclusion: This study demonstrated statistically significant differences between physostigmine and placebo on both a performance-based cognitive functioning instrument and a clinician's global evaluation. The magnitude of the effect size was small and occurred only in the subset of patients who responded in the initial dose titration study period. Nevertheless, the results suggest that in a subset of patients, physostigmine can induce a degree of cognitive improvement over 6 weeks of treatment.