OBJECTIVE Epidemiologic evidence for a protective effect of exogenous female sex hormones on the development of rheumatoid arthritis (RA) is contradictory. We examined whether exposure to either oral contraceptives (OC) or postmenopausal estrogen replacement therapy (ERT) is associated with the development of RA in women. METHODS We separately examined the relationship between use of OC and ERT on the risk of RA in a population based case-control study. Case patients, including all female residents of Rochester, Minnesota, > or = 18 years of age, who first fulfilled 1987 American College of Rheumatology criteria for RA between 1955 and 1994 (n = 445), were compared with age matched female controls from the community. Multivariable conditional logistic regression models were used to determine whether OC or ERT exposure had an effect on RA development after controlling for potential confounders. RESULTS We observed an inverse association between ever-use of OC and the risk of RA, which persisted after adjusting for potential confounders in multivariate analyses (OR 0.56, 95% CI 0.34, 0.92). Earlier calendar-year of first exposure to OC was associated with lower OR for RA. We found no evidence of a significant association of ERT with RA risk (adjusted OR 1.11, 95% CI 0.69, 1.78). CONCLUSIONS Exposure to OC, but not ERT, significantly reduces the risk of development of RA. The risk of developing RA is lower when OC exposure occurred in earlier years, which suggests that the higher doses of estrogens and progestins contained in earlier OC preparations may have a stronger protective effect against developing RA. While this protective effect is strong, it only explains a small portion of the observed decrease in RA incidence over the past few decades because the proportion of Rochester women exposed to OC is quite small.