Summary Among patients with gastric cancer (GC) and gastroesophageal cancer (G-EC), HER2 amplification identifies those who may benefit from trastuzumab. HER2 status assessment, however, is influenced by preanalytic, analytic, and postanalytic variables. In a series of 5426 microarray cancer tissue cores obtained from 1040 GC/G-ECs (824 GC, 216 G-EC) and 720 synchronous nodal metastases, we evaluated both the performances of 2 different immunohistochemistry (IHC) protocols and the HER2 status intratumor variability. The prevalence of HER2 amplification and protein overexpression were assessed by chromogenic in situ hybridization and by 2 IHC protocols (CB11 and 4B5). HER2 was amplified in 114 (11%) of 1040 cases; in 6 (5.3%) of 114 cases, gene amplification only involved nodal metastasis. HER2 amplification prevailed in intestinal-type ( P = .001) and low-grade ( P < .001) tumors, showing no correlation with patients’ age/sex, tumor location, stage, and Ming histotype. Overall, 12.5% and 13.7% of cases IHC scored 2+/3+ using the CB11-IHC and the 4B5-IHC protocol, respectively. HER2 amplification was not associated with protein overexpression (score 0/1+) in 11.4% and 6.2% of cases using the CB11-IHC and the 4B5-IHC protocol, respectively. The 4B5-IHC protocol proved more sensitive than CB11-IHC (93.9% versus 88.6%) and just as specific (96.1% versus 96.9%). Tested by chromogenic in situ hybridization, intratumor HER2 status was “substantially” consistent in different tissue cores obtained from the same case ( κ = 0.78). Similar results were obtained for HER2 protein expression (CB11-IHC, κ = 0.78, and 4B5-IHC, κ = 0.83). Immunohistochemistry testing, however, fails in identifying about 10% of HER2 -amplified cancers, potentially excluding these patients from anti-HER2 therapy.