Intraperitoneal injection of physostigmine in rats produced a retrograde amnesia of a trained task of escaping shock. This amnesic effect was a U-shaped function of the length of the interval between initial training and injection. In all cases, retraining Occurred 30 minutes after injection. A substantial effect was produced by physostigmine if its application was made 30 minutes after training; there was no effect if application and tests were made 1, 2, or 3 days after the original training. When the substance was injected and the rats were retrained 5, 7, or 14 days after the original training, a substantial effect again appeared. These results are similar to those reported in experiments in which another anticholinesterase, diisopropyl fluorophosphate, was applied intracerebrally. The data demonstrate a similar pattern of change of the amnesia with time, and they substantiate the view that neither the place of application nor the brain lesions caused the reported amnesia.