Conflicting data have been reported about the association between glutathione S-transferase (GST), a family of proteins implicated in detoxification of cytotoxic drugs in human ovarian in vitro models, and response to chemotherapy and prognosis in ovarian cancer patients. The aim of this study was to analyze the possible clinical role of GST activity in a large series of primary ovarian cancer patients. The study included a large series of primary untreated ovarian cancer patients who underwent cytoreductive surgery and chemotherapy and who were followed up in a single institution. GST activity levels were assessed in tumor extracts by using a biochemical assay. A cut-off of 250 units of enzymatic activity was chosen according to the receiver operating characteristics (ROC) curve. GST activity levels were distributed in an asymmetrical manner (median: 266 units; range: 4-918 units) and did not seem to be associated with stage, histopathological grading, ascites, or residual tumor after surgery. Higher GST activity levels were found in patients who responded to chemotherapy (median: 298 units, range: 50-691) than in those who responded only partially (median: 227 units, range: 19-747) or not at all to chemotherapy (median: 246 units, range: 4-811) (H = 7.02, P = 0.029). Moreover, the percentage of cases with > 250 units was significantly higher among complete responders (66%) than partial responders (37%) or non-responders (48%) (chi 2 = 7.32; P = 0.025). When multivariate analysis, including clinico-pathological parameters and GST activity status as predictors of response to chemotherapy, was carried out, residual tumor, stage and GST status retained independent predictive value. Patients with high GST activity had more favourable prognosis than those with low GST activity. The median PFS was 42 months for patients with high GST activity compared to 17 months for those with low GST activity (P = 0.037). The median overall survival was 72 months for high-GST-activity and 42 months for low-GST-activity patients (P = 0.043). Substantially similar results were obtained in the subgroup of stage II-III-IV ovarian cancer patients. Multivariate analysis including the clinico-pathological parameters and GST activity status was performed in stage III-IV ovarian cancer patients: Stage IV disease, residual tumor > 2 cm, the presence of ascites and low GST activity status retained independent negative prognostic roles. A direct association between high GST activity and a better clinical outcome in terms of response to chemotherapy and survival has been observed in a large series of primary untreated ovarian cancer patients. These results, which are contrary to the expectations raised by in vitro studies, emphasize the need for caution when translating in vitro-generated hypotheses to the clinical setting.