A review of the literature concerning cholinergic manipulations and passive avoidance reveals that state-dependency is usually not controlled adequately, nor is acquisition of the passive avoidance task ascertained before retention is tested. These problems make interpretation of results difficult. We report three experiments on 129 rats, controlling both of these factors, in which scopolamine and pilocarpine impaired both acquisition and retention of a passive avoidance response. Lesions of the nucleus basalis magnocellularis had no effect on this task. The results suggest that an optimal level of central cholinergic activity exists for learning and memory, and that deviations from this optimal level impair acquisition and retention.