The vaginal douching habits of 100 consecutive municipal hospital patients with verified pelvic inflammatory disease (cases) were compared with those of 762 randomly selected controls (random controls) and 119 women thought to have pelvic inflammatory disease but in whom the diagnosis was not confirmed by laparoscopy and/or endometrial biopsy specimen (internal controls). Because patients had been symptomatic for no more than 3 weeks, current douching was arbitrarily defined as any douching during the previous 2 months. Current douching was more common among those with pelvic inflammatory disease than among random controls or internal controls. Among current douchers, pelvic inflammatory disease was significantly related to frequency of douching. For example, when cases were compared with random controls, those who douched three or more times per month were 3.6 times more likely than those who douched less than once per month to have confirmed pelvic inflammatory disease. A logistic regression model was used to adjust for demographic, behavioral, and other possible confounding variables. Even after adjustments, douching during the previous 2 months remained associated with pelvic inflammatory disease. These data suggest that among these women vaginal douching may be a risk factor for pelvic inflammatory disease.