Patients with uncomplicated pelvic inflammatory disease (PID) (acute salpingitis and no pelvic masses) were randomly assigned for treatment with either cefotaxime or cefoxitin. A clinical cure was achieved in 17 of 20 cases (82%) and 19 of 22 cases (84%), respectively. Within the complicated PID group, patients were assigned to two subgroups: those with a tubo-ovarian complex (26 patients), and those with a tubo-ovarian abscess (32 patients), as confirmed by ultrasonography or surgery. Patients within each of these two subgroups were then randomly assigned for treatment with either cefotaxime or clindamycin plus gentamicin. Within the tubo-ovarian complex subgroup, a clinical cure was achieved in 11 of 13 cases (85%) treated with cefotaxime and 10 of 13 cases (77%) treated with clindamycin plus gentamicin. Within the tubo-ovarian abscess subgroup, a clinical cure was achieved in 12 of 16 cases (75%) treated with cefotaxime and 11 of 16 cases (69%) treated with clindamycin plus gentamicin. No differences in any category were statistically significant. Specimens for culture were obtained from the endocervix, endometrium, and when possible, the cul-de-sac, fallopian tubes, and abscess. Neisseria gonorrhoeae (33%) was isolated more frequently than Chlamydia trachomatis (12%) in patients with PID, and neither of these organisms was isolated with any increased frequency in patients with complicated PID. The majority of the patients were considered to have polymicrobial infection. Cefotaxime was as efficacious as cefoxitin and clindamycin plus gentamicin for the treatment of acute salpingitis, tubo-ovarian complex and tubo-ovarian abscess.