Trastuzumab combined with chemotherapy is standard of care for HER2 positive advanced gastro-esophageal cancers. The reported prevalence of HER2 discordance between primary tumors and corresponding metastases varies, hampering uniform patient selection for HER2 targeted therapy. This meta-analysis explores the influence of HER2 assessment methods on this discordance and investigates the prevalence of HER2 discordance in gastro-esophageal adenocarcinomas. PubMed, Embase and Cochrane databases were searched until January 2016. Differences in discordance rate between strict and broad(er) definitions of HER2 status were assessed using random-effect pair-wise meta-analysis. Random-effect single-arm meta-analyses were performed to assess HER2 discordance and the prevalence of positive and negative conversion. A significantly lower discordance rate in HER2 status between primary tumors and corresponding metastases was observed using a strict vs. broad definition of HER2 status (RR = 0.58, 95%CI 0.41–0.82), with a pooled discordance rate of 6.2% and 12.2%, respectively. Using the strict definition of HER2 assessment pooled overall discordance was 7% (95%CI 5–10%). The lowest discordance rates between primary tumors and corresponding metastasis are observed when using a strict method of HER2 positivity. Treatment outcomes of different studies will be better comparable if selection of eligible patients for HER2 targeted therapy is based on this strict definition.