OBJECTIVES: To estimate the relative treatment effects on progression free survival (PFS) and overall survival (OS) of trastuzumab plus an aromatase inhibitor (AI), lapatinib plus an AI, and AI monotherapy (anastrozole, letrozole or exemestane) in postmenopausal women with untreated metastatic hormone receptor positive breast cancer which over-expresses HER2. METHODS: A systematic literature review was conducted in MEDLINE, MEDLINE-IN-PROCESS, EMBASE, Cochrane CENTRAL, Web of Science Proceedings and BIOSIS to retrieve relevant randomised clinical trials. Heterogeneity and inconsistency were discussed with clinical experts and assessed statistically. An MTC was developed within the Bayesian framework using non-informative priors. It was decided a priori to investigate whether AIs could be considered as a class, through the evaluation of hazard ratios between single agents. As few trials focused on the exact target population, extensive sensitivity analyses were planned. RESULTS: The base case analysis indicated no significant difference between AIs on PFS. Combined therapies were significantly more effective than AIs: the hazard ratios were estimated at 0.55 (95%CrI: [0.41; 0.74]) for trastuzumab+AI and 0.71 (95%CrI: [0.53; 0.96]) for lapatinib+AI. Although the probability of being best varied in the sensitivity analyses, trastuzumab+AI was always associated with the highest probability of best efficacy. The effect on OS was similar across AIs. No significant difference was found between the two combination therapies, which were associated with high probabilities of best efficacy (48% for trastuzumab+AI and 50% for lapatinib+AI). The sensitivity analyses indicated that adjustment for cross-over had an important impact on the OS results. CONCLUSIONS: Our study confirmed that anastrozole, letrozole and exemestane had similar efficacy with respect to PFS and OS and that combined therapies were associated with significantly improved PFS compared to AIs. Conducting an MTC in this specific population was addressed through clinical and statistical assessments of heterogeneity and inconsistency, and extensive sensitivity analyses.