Seventy volunteers were injected with diazepam (0.3 mg/kg), scopolamine (8 mug/kg), or placebo, followed 70 min later by another injection of physostigmine, physostigmine and methscopolamine (in case of diazepam treatment), or placebo. Physostigmine was given in two doses, 16 and 32 mug/kg; methscopolamine, 8 and 16 mug/kg. Subjects (Ss) were tested in groups of 5 in a double blind procedure with treatments distributed according to a Latin square design. Prior to treatment, Ss heard a series of lists of words, followed by an immediate recall test. Following the first injection, delayed free recall and recognition tests were given. The second drug was then injected, followed by a presentation of another two sets of lists which were tested similarly. Subjective feelings were also evaluated with a rating questionnaire. Diazepam and scopolamine did not affect recall of information which had been learned prior to drug injection. However, both drugs impaired the learning or acquisition of new information. Physostigmine, especially in its high dose, antagonized most of the memory deficits produced by scopolamine while those of diazepam remained. This is a strong indication that scopolamine acts centrally through an anticholinergic mechanism while diazepam may act through a different system.