Measurements of prepregnant body weight have important research and clinical applications. In practice, however, they are not always recorded; even when they are, this information is not always readily available. For this reason, researchers and clinicians have to rely on retrospective estimates of prepregnant weight, which can be estimated using: 1) maternal self-reports, 2) retrospective extrapolation, or 3) standardized estimates that correct for weight gained during early pregnancy. The aim of the present study was to examine the relative merits of these three approaches. Maternal self-reports tend to be unreliable and biased, being influenced by a variety of sociodemographic characteristics that generally underestimate true prepregnant body weight. Estimates of prepregnant weight based on retrospective extrapolation are vulnerable to measurement error, transient fluctuations in body weight, and incorrectly assume that the rate of weight gain is constant throughout pregnancy. Standardized estimates that correct for weight gained during early pregnancy incorrectly presume that there is little interindividual variation in gestational weight gain and that weight gain is similar for each woman in consecutive pregnancies. Because none of these techniques can provide a precise measure of prepregnant weight, researchers have little alternative but to recruit and weigh women before they become pregnant, although measurements of body weight recorded during the first trimester of pregnancy may provide a reasonable indication of prepregnant weight. For clinicians, self-reports of prepregnant weight or measurements recorded early in pregnancy are probably sufficiently accurate for practical purposes whenever recent, accurate measurements of prepregnant weight are unavailable.