The estrogen receptor (ER) is the target of tamoxifen, but endocrine therapies do not benefit all patients with ER positive tumors. We therefore hypothesized that copy number changes in the ESR1 gene, encoding ER, confer resistance. Within a consecutive series of ER positive, postmenopausal patients allocated to 5 years tamoxifen, we identified 61 patients with recurrence less than 4 years and 48 patients without recurrence at least 7 years after initiation of adjuvant tamoxifen. Archival tissue containing primary tumor was collected from 97 patients (89%). Tumor samples were analyzed for ESR1 copy number changes using FISH with a probe covering the ESR1 gene at 6q25 and a reference probe covering the centromere of chromosome 6. The assay was validated in a material of 120 normal breast samples. FISH analysis for ESR1 was successful in 91 patients (94%). Amplification (ratio ESR1/CEN-6 ≥ 2.0) was observed in 11 of 50 (22%) patients with early recurrence, compared to two of 41 (5%) patients without recurrence. The difference is statistically significant (P = 0.033). In both groups, two patients with ESR1 deletion (ratio ESR1/CEN-6 < 0.8) were identified. ESR1 amplification was significantly associated with poor disease-free survival (P = 0.0054) and overall survival (P = 0.0004). This pilot study supports our hypothesis that ESR1 amplification is associated with a poorer outcome following adjuvant treatment with tamoxifen in ER positive early breast cancer. This study also revealed the existence of ESR1 deletions. The prognostic and predictive impact of ESR1 copy number changes needs further exploration in clinical trials.