This paper discusses the importance of specification of risk periods in research of the effect of pregnancy on the occurrence of gallstone disease. The authors present data from a case-control study, started in 1983 in Maastricht, The Netherlands. The study comprised 100 acute gallstone cases and 305 general population controls and 41 elective cases and 122 elective radiodiagnosis controls. In a conventional analysis, the rate ratio for one or more pregnancies ever (vs. never) was 1.2 (90% confidence interval 1.02-1.36). In an analysis that discerned several subsequent risk periods, an effect was found only for the risk period up to 5 years after pregnancy (rate ratio, 2.4; 90% confidence interval 1.12-4.96). Beyond this risk period no effect was noted. The results suggest that pregnancy increases the rate of gallstone formation only transiently. Six reasons for specifying risk periods in nonexperimental etiologic studies are discussed. The authors contend that ignoring of risk periods explains part of the variation in the findings of earlier studies on pregnancy and the risk of gallstone disease.