This paper presents a summary of the evidence review group (ERG) report into trastuzumab for the treatment of human epidermal growth factor receptor 2 (HER2)-positive metastatic adenocarcinoma of the stomach (mGC) or gastro-oesophageal junction. HER2 positivity is defined by immunohistochemistry (IHC)3+ or IHC2+/fluorescence in situ hybridisation (FISH)+. The decision problem addressed was the testing of the whole mGC population with IHC and, for IHC2+ patients, also with FISH, followed by treatment of HER2-positive patients with trastuzumab combined with cisplatin and either capecitabine or 5-fluorouracil (5-FU) [HCX (trastuzumab, cisplatin, capecitabine)/fluorouracil (F)] compared with current standard NHS therapy. The manufacturer's submission contained direct evidence from the ToGA trial, a well-conducted, multinational, phase III randomised controlled trial (RCT) that compared HCX/F with cisplatin and a fluoropyrimidine alone [cisplatin, capecitabine (CX)/F]. HCX/F showed statistically significantly better overall survival in the European Medicines Agency-licensed population subgroup (74%) (hazard ratio 0.65, 95% confidence interval 0.51 to 0.83), corresponding to median survival of 16 months versus 11.8 months. No other evidence exists for the efficacy of any therapy in a known HER2-positive mGC population; other comparisons extrapolate from trials in mixed HER2 status populations. The ERG accepted the manufacturer's view that a meaningful network meta-analysis to establish a comparison for HCX/F compared with current standard NHS therapy [epirubicin, cisplatin, capecitabine (ECX)/epirubicin, oxaliplatin, capecitabine (EOX)/epirubicin, cisplatin, 5-FU (ECF)] was not possible, but was unconvinced by arguments advanced in the alternative narrative synthesis. These involved disregarding evidence from a meta-analysis and interpreting non-significant results of small RCTs comparing epirubicin-containing triplets with cisplatin, 5-FU (CF)/capecitabine (X) doublets as evidence of no difference between triplet and doublet regimens. The high CX/F dose in the ToGA trial was an additional basis for the contention of equivalence. An appropriate de novo economic evaluation, including an economic model that separately compared HCX or trastuzumab, cisplatin, 5-FU (HCF) with the triplet regimens ECX, EOX and ECF, based on a simple, three-state cohort model (progression-free, disease, progression and death), was submitted. Utility weights were applied to estimate quality-adjusted life-years (QALYs). Costs were assessed from an NHS perspective, and incorporated the acquisition and monitoring costs of the alternative regimens, HER2 testing, adverse events and other supportive care costs. An 8-year time horizon was used to represent a lifetime analysis. Results from the ToGA trial were combined with a series of assumptions on relative treatment effects and testing strategies. The manufacturer's results produced an incremental cost-effectiveness ratio (ICER) of £ 53,010 per QALY for HCX versus ECX. Although the manufacturer undertook a detailed set of sensitivity analyses, several alternative model assumptions were not evaluated. The ERG undertook a series of alternative base-case analyses. As a result of these analyses, EOX replaced ECX as the appropriate comparator, and the ICER for the comparison of HCX vs EOX increased to between £ 66,982 and £ 71,636 per QALY. The impact of implementation of alternative testing strategies remained unclear. There is also considerable uncertainty surrounding the true estimate of effectiveness for the comparison between triplet regimens containing epirubicin (ECX/ECF/EOX) and doublet CX/F regimens. Consequently, the view of the ERG was that there is insufficient evidence on the efficacy of HCX/F compared with current NHS standard therapy for an ICER to be determined with any degree of certainty.