The possibilities and limitations of the pharmacoepidemiologic methods of determining neuropharmacologic drug efficacy and safety in psychiatric patients are discussed in this review. Such methods can improve both the scientific evaluation and clinical practice aspects of our knowledge of these drugs by providing: (1) incidence rates of drug usage and adverse events from computerized information on large populations; (2) quantitative methods for risk-to-benefit assessments that incorporate multiple outcome measures, provide long-term effectiveness and safety data, and use statistical methods to distinguish drug-induced from illness-based behaviors; and (3) systematic epidemiologic approaches to resolving dilemmas that involve the political and social context in which neuropharmacological drugs are used.