The purpose of this study was to estimate in all randomised trials the relative risk of overall response rate (ORR), clinical benefit (CB), time to progression (TTP), overall survival (OS), and toxicity of aromatase inhibitors (AI), compared with tamoxifen (Tam) as first-line endocrine therapy in postmenopausal metastatic breast cancer (PMBC) women. Prospective randomised studies were searched through computerised queries of MEDLINE, EMBASE, and the American Society of Clinical Oncology (ASCO) abstract database. Relative risk, 95% confidence interval, and heterogeneity were derived according to the inverse variance and Mantel-Haenszel method and Q statistics. Six phase III prospective randomised trials including 2787 women were gathered. A significant advantage in ORR (P = 0.042), TTP (P = 0.007), and CB (P = 0.001) in favour of AI over Tam was detected at the fixed effects model. These results were not significant at the random effects model, owing to the significant heterogeneity. On the contrary, no difference was registered for OS (P = 0.743) with no significant heterogeneity. Regarding toxicity, Tam caused more frequently thromboembolic events (P = 0.005) and vaginal bleeding (P = 0.001) compared with AI. Aromatase inhibitors appear to be superior to Tam as first-line endocrine option in PMBC women. Owing to a component of variability between the six studies analysed, the random effects estimates differed from corresponding fixed ones. Investigators should assess heterogeneity of trial results before deriving summary estimates of treatment effect.