Corynebacterium parvum was administered intraperitoneally to 21 patients with epithelial ovarian cancer. Nineteen patients had surgically measurable disease and two received adjuvant therapy. Surgically confirmed responses were documented in six of 19 patients (31.6%), with two complete responses (10.5%) and four partial responses (21.1%). Three patients (15.8%) had stable disease, and 10 patients (52.6%) had disease progression. The mean survival of the patients who had a complete response was 35.5 months; the four patients who had a partial response the mean survival was 26.6 months, and of the nonresponders the mean survival was 12.6 months (p less than 0.02). The mean survival of the entire group was 18.2 months. Initial response and patient survival correlated with the amount of disease pretreatment. Thus six responding patients had less than or equal to 5 mm maximum diameter tumors, that is, minimal residual disease. Toxicity in the 86 courses of therapy included abdominal pain in 78% of cases, fever in 56%, nausea in 40%, and vomiting in 22%. Stimulation of cytotoxic lymphocytes resulted from the administration of C. parvum, which induced a significant increase of both intraperitoneal natural killer lymphocyte cytotoxicity and antibody-dependent cell-mediated cytotoxicity in six of nine patients tested; these two types of cytotoxicity correlated with response to therapy and may be partially responsible for the surgically documented tumor regression. While the clinical usefulness of intraperitoneal C. parvum is limited because of its toxicity, intraperitoneal immunotherapy may prove useful in patients with minimal residual ovarian cancer when more refined agents become available.