The design of cohort and case/control studies to assess potential reproduction effects of toxic agents is reviewed. Factors of importance in the design of cohort studies of occupational exposures include the number of events needed to examine outcomes of interest, probable number of events identified in a given population reflecting sociodemographic characteristics of the work force, ability to characterize exposures, potential data sources, and selection of appropriate statistics. A cohort of sufficient size to examine a spectrum of reproductive outcomes, with good quantitative and qualitative data, is the preferred method of evaluating reproductive effects of exposures to potentially toxic agents. Industry based studies are typically retrospective, and the current age distribution of the population effects the actual time span covered by the study. When female workers are studied, inclusion of terminated workers is essential to prevent biased assessment of certain pregnancy outcomes, and demographic, medical, and occupational characteristics of the study population. Case/control studies associating occupational exposures with particular reproductive outcomes can be drawn from a large cohort study of an industrial population, or may be drawn from community medical or vital records. One serious problem is the potential for under reporting of outcomes and the potential for bias. Assessment of pregnancy outcomes identified in hospital, physician, or vital statistics records used in case/control studies in the general population. The author concludes that the two primary factors to be considered in the design of pregnancy outcome studies are the continuum of outcomes that may be observed and their relationship to degree of exposure.