Twelve patients aged 53 to 89 years with presenile or senile dementia of the Alzheimer's type participated in a treatment trial of oral physostigmine. Patients first received the drug during an open dose-finding phase. In a double-blind crossover design phase, the patients' memory performance during treatment with an optimal memory-enhancing dose was compared with memory performance during placebo treatment. Response to medication was assessed by the Delayed Recognition Span Test, the Selective Reminding Test, and the Alzheimer's Disease Assessment Scale. Intergroup differences in response to physostigmine were nonsignificant. Age at onset of the disease did not differentiate partial responders from nonresponders.