Clinical trials implicate the estrogen receptor (ER) coactivator amplified in breast cancer 1 (AIB1) to be a prognostic and a treatment-predictive factor, although results are not unanimous. We have further investigated this using a controlled randomised trial of tamoxifen versus control.A total of 564 premenopausal women were entered into a randomised study independent of ER status. Using a tissue microarray, AIB1 and ER were analysed by immunohistochemistry.AIB1 scores were obtained from 349 women. High AIB1 correlated to factors of worse prognosis (human epidermal growth factor receptor 2, Nottingham histological grade 3, and lymph node metastases) and to ER negativity. In the control arm, high AIB1 was a negative prognostic factor for recurrence-free survival (RFS) (P = 0.02). However, ER-positive patients with high AIB1 responded significantly to tamoxifen treatment (P = 0.002), increasing RFS to the same level as for systemically untreated patients with low AIB1. Although ER-positive patients with low AIB1 had a better RFS from the beginning, this was not further improved by tamoxifen (P = 0.8).In the control group, high AIB1 was a negative prognostic factor. However, ER-positive patients with high AIB1 responded significantly to tamoxifen. This implicates high AIB1 to be an independent predictive factor of improved response to tamoxifen and not, as has previously been discussed, a factor predicting tamoxifen resistance.