HER2 protein overexpression and gene amplification are important biomarkers for identifying gastric cancer patients who may respond to HER2-targeted therapy using trastuzumab. The aim of this study was to evaluate the concordance between HER2 protein expression and gene amplification in both surgically resected tumors and matched biopsy specimens of gastric cancer. Formalin-fixed, paraffin-embedded sections of 207 surgically resected tumors and 158 biopsy specimens from 207 cases of invasive intestinal-type gastric cancer were analyzed. Protein expression was assessed using immunohistochemistry and graded by the modified scoring criteria for gastric cancer. Gene amplification was evaluated by fluorescence in situ hybridization (FISH). HER2 overexpression was observed in 17 % of both surgically resected tumors (35/207) and biopsy specimens (26/158). HER2 gene amplification was detected in 31 % (61/200) of surgically resected tumors and 32 % (47/147) of biopsy specimens. Except for immunohistochemistry (IHC) equivocal (2+) cases, the concordance rates between IHC and FISH was 90.9 % in surgically resected tumors and 90.2 % in biopsy specimens. In IHC 2+ cases, the rate of HER2 gene amplification was 56 and 38 % in surgically resected tumors and biopsy specimens, respectively. IHC-FISH discordance was mainly due to intratumoral heterogeneity and low-level gene amplification. The concordance rate of IHC results between surgically resected specimens and the corresponding biopsy specimen was 57.0 % (κ = 0.224), and in discordant cases, HER2 positivity in biopsies and HER2 negativity in surgically resected tumors were most common. The concordance rate of FISH results between surgically resected tumors and biopsy specimens was 72.7 % (κ = 0.313). Polysomy 17 was detected in 5.5 and 7.5 % of surgically resected tumors and biopsy specimens and significantly correlated with IHC score, but polysomy 17 could explain one IHC score 3+ and FISH-negative tumor only. Although high concordance rates between HER2-protein expression and gene amplification were observed in both surgically resected tumors and biopsy specimens, the agreement levels were evaluated to be fair. Polysomy 17 was infrequent and seemed to have limited impact on gastric HER2 testing. Further investigations are required for an appropriate biopsy method to reduce false results of HER2 testing and to clarify the clinical significance of intratumoral heterogeneity in HER2 status.