Amplification for the ERBB2 oncogene encoding the HER2/neu protein (HER2) is of predictive and prognostic importance in breast carcinoma. Fluorescence in situ hybridization (FISH) is a widely accepted method for determining HER2 amplification status. A HER2-amplified tumor is defined as having a ratio of HER2 signals to chromosome 17 centromeric probe signals (HER2/CEP17 ratio) exceeding 2.2. However, the presence of scattered cells demonstrating HER2 amplification is of unclear significance. A 2009 panel guideline defined a tumor with 'genetic heterogeneity' as having at least 5% but fewer than 50% of (non-clustered) tumor nuclei with a ratio >2.2. The study objective was to examine the statistical distribution of breast tumors tested by FISH for HER2 amplification, after implementation of this 2009 guideline. We identified 2522 consecutive breast carcinoma cases (2009-2011) tested for HER2 amplification. All cases were tested by FISH using a standard clinical protocol, adhering to established guidelines. For each case, data on cell counts were retrieved electronically. Each tumor was compared with a theoretical normal distribution by quantile-quantile analysis. Of 2522 FISH tests for HER2, 1900 (75%) were non-amplified, 394 (16%) were amplified, and 228 (9%) were HER2-equivocal. A total of 666 (26%) had 'genetic heterogeneity.' Among these 'genetically heterogeneous' cases, the ratio was non-amplified in 430 (64.5%), amplified in 24 (4%), and equivocal in 212 (31.5%). The amplified subpopulation in 'genetically heterogeneous' tumors was larger if the overall ratio was close to 2.2. However, the percentage of nuclei >2.2 in a 'genetically heterogeneous' tumor was not informative of the underlying tumor-cell distribution. We conclude that the proportion of HER2-amplified nuclei within a tumor does not contribute information independent of the actual HER2/CEP17 ratio. Reassessment of the definition of 'genetic heterogeneity' in HER2 testing is warranted.