The protective effect of barrier-method contraception against pelvic inflammatory disease (PID) was examined by analyzing data from the Women's Health Study, a large multicenter case-control study. We compared the contraceptive methods used by 645 women hospitalized for initial episodes of PID with the contraceptive methods used by 2,509 control subjects reporting no history of PID. The risk of hospitalization for PID in women currently using barrier methods relative to women using all other methods and to women using no method of contraception was 0.6 (95% confidence limits, 0.5 to 0.9) for both comparisons. This protective effect was observed for both chemical and mechanical barrier methods, although it was not statistically significant for the former. The prevention of PID and its sequelae is one of the most important noncontraceptive benefits of barrier methods of contraception.