Background: P-glycoprotein (gp170; encoded by the MDR1 gene [also known as PGY1]) is a membrane protein capable of exporting a variety of anticancer drugs from cells. MDR1/gp170 expression has been studied in breast cancer, but the prevalence of this expression and its role in breast tumor drug resistance are unclear. Purpose: We conducted a critical review and meta- analysis of studies examining MDR1/gp170 expression in breast cancer to estimate the likely prevalence and clinical relevance of this expression. We also explored reasons for differences in the findings from individual studies. Methods: Published papers on MDR1/gp170 expression in breast cancer were identified by searching several literature databases and reviewing the bibliographies of identified papers. Variability across the studies in the proportion of tumors expressing MDR1/gp170 was assessed by use of chi- squared tests of homogeneity, weighted means, and weighted linear regression. Pooled relative risks (RRs) for the association between the induction of MDR1/gp170 expression and prior chemotherapy and associations between MDR1/gp170 expression and several clinical outcomes were estimated by use of Mantel-Haenszel methods. Heterogeneity among the pooled RRs was explored by use of chi-squared tests. Reported P values are two-sided. Results: Thirty- one studies were identified and evaluated. The proportion of breast tumors expressing MDR1/gp170 in all of the studies was 41.2%, but there was substantial heterogeneity in the values across individual studies (P<.0001). Regression analyses demonstrated that a considerable portion of the observed heterogeneity was a consequence of the change, over time, from RNA hybridization-based assays to immunohistochemistry-based assays of MDR1/gp170 expression. Measuring MDR1/gp170 expression before versus after chemotherapy and use of cytotoxic drugs that are not substrates for gp170 also contributed to the heterogeneity. Treatment with chemotherapeutic drugs or hormonal agents was associated with an increase in the proportion of tumors expressing MDR1/gp170 (RR = 1.77; 95% confidence interval [CI] = 1.46-2.15). Patients with tumors expressing MDR1/gp170 were three times more likely to fail to respond to chemotherapy than patients whose tumors were MDR1/gp170 negative (RR = 3.21; 95% CI = 2.28-4.51); this RR increased to 4.19 (95% CI = 2.71- 6.47) when considering only patients whose tumor expression of MDR1/gp170 was measured after chemotherapy. MDR1/gp170 expression was not associated with lymph node metastases, estrogen receptor status, tumor size, tumor grade, or tumor histology. Conclusions and Implications: MDR1/gp170 expression in breast tumors is associated with treatment and with a poor response to chemotherapy. The data are consistent with a contributory role for MDR1/gp170 in the multidrug resistance in some breast tumors.