Evidence-based medicine that is designed to guide benefit/risk drug therapy decisions does not exist for pregnant women. The types of studies that do exist are usually conducted in animals, which may not reflect human benefits and risks. The types of studies that do exist in humans are typically limited and, at best, may show an "association" between a particular drug therapy and an undesirable effect. This review outlines the difficulties that are associated with the assessment of the benefits/risks of drug therapy during pregnancy, the history of the Food and Drug Administration regulations for labeling prescription drugs, and the strengths and weaknesses of the current Food and Drug Administration pregnancy labeling system. Proposed changes to the current system are reviewed.