OBJECTIVE We evaluated the relationship between clinically severe pelvic inflammatory disease and laparoscopic diagnosis and grading, comparative treatment with clindamycin plus cefamandole or doxycycline, and a management protocol for inpatient pelvic inflammatory disease treatment. METHODS Thirty-three patients who met our clinical criteria for severe pelvic inflammatory disease underwent diagnostic laparoscopy. Pelvic inflammatory disease patients were randomized to double-blind treatment with clindamycin plus cefamandole or doxycycline within our management protocol; postdischarge oral antibiotics were omitted. RESULTS Laparoscopy confirmed pelvic inflammatory disease in 23 (70%) patients; 10 (44%) had mild pelvic inflammatory disease by laparoscopic grading. Laparoscopic grade alone predicted necessary duration of therapy to response: mild pelvic inflammatory disease, 2.3 +/- 0.5 days; moderate pelvic inflammatory disease, 2.7 +/- 1.5 days; and severe pelvic inflammatory disease, 3.9 +/- 1.5 days (p less than 0.05). Using the management plan presented, response rates for both antibiotic regimens were 100%. CONCLUSIONS Clinical diagnosis and grading of severe pelvic inflammatory disease has poor specificity. Laparoscopic grading of severity of pelvic inflammatory disease seems accurate. Both clindamycin plus cefamandole and clindamycin plus doxycycline are equally effective regimens for treatment of pelvic inflammatory disease and did not require supplementation after discharge. Our management plan is objective and practical; daily bimanual examination is the most sensitive indicator of persistent disease.