BACKGROUND--Although more epidemiological studies of reproductive health have been published recently, important scientific achievements remain few. Among the reasons for this are existing methodological constraints. Fertility is low in most parts of the industralized world, making the observation time short. Pregnancy itself is under extremely close medical surveillance with intensive health care interventions such as prenatal screening, induced abortions, induced labour, and prescribed changes in environmental exposures, but the epidemiologists are generally left with cause-effect experiences which escaped medical attention or study of diseases for which there is no cure. The main problem in non-experimental research of reproductive failure, however, is the most pregnancies are planned and past pregnancy experience is used in the planning. Self selection into different exposure categories according to past pregnancy experience should therefore be considered when studies are designed and analysed. STUDY OBJECTIVE--To consider how to deal with past pregnancy experience as an indicator "background" risk of reproductive failure. The present standard is to take parity or gravidity into consideration but this may be of little value and could even be very misleading. MAIN RESULTS--No simple and universal analytic strategy is available in reproductive epidemiology. On some occasions studies should be restricted to the first pregnancy only, in other situations the past reproductive experience may be used to form groups that are comparable in terms of pre-exposure risk.