Glutathione S-transferase pi (GST-pi), a Phase II detoxification enzyme, has recently been implicated in protection against apoptosis. Expression of GST-pi and Bcl-2 protein, an established apoptosis marker, was analyzed by immunohistochemistry in 116 cases of infiltrative ductal breast carcinomas in Singapore women. The markers were correlated with apoptosis detected by the TUNEL method and clinico-pathological parameters. There were 67 (58%) GST-pi-positive breast tumors and 43 (37%) Bcl-2-positive tumors. In a large proportion of GSTpi-positive/Bcl-2-positive tumors, there was a distinct accumulation of the GST-pi enzyme within the nucleus of cancer cells when examined by double immunofluorescence labeling under confocal microscopy. GST-pi immunoreactivity was not significantly correlated with any of the traditional histologic factors known to influence prognosis, whereas Bcl-2 overexpression was associated with reduced size of primary tumor (P ؍ .021) and positive estrogen receptor status (P ؍ .001). Univariate analysis revealed that GST-pi-positive, Bcl-2-positive, and lower histological grade tumors had decreased levels of apoptosis (P ؍ .024, P ؍ .011, and P ؍ .029, respectively). However, multivariate analysis showed that histological grade and Bcl-2, but not GST-pi, immunoreactivity were correlated with apoptotic status. The Kaplan-Meier disease-free survival curves showed a significant difference between GST-pi-positive and GST-pi-negative breast cancer cases (P ؍ .002). Disease-free survival in patients with GST-pi-positive tumors was also worse than that in patients with GST-pi-negative tumors in the group who had adjuvant chemotherapy (P ؍ .04). In patients who were lymph node positive, GST-pi immunopositivity was found to influence disease-free survival. Recurrence of tumors was also significantly affected by GST-pi immunoreactivity (relative risk of 8.1). The findings indicate that GST-pipositive tumors are more aggressive and have a poorer prognosis than do corresponding GST-pinegative breast cancers.