We have analysed the pedigrees of all 70 women who developed cancer in the Royal Marsden Hospital (RMH) tamoxifen chemoprevention trial, using the Claus model, to assess breast cancer susceptibility heterozygote risk (HR) and screened the entire coding regions of BRCA1 and 2 genes in 62 of these cases. We found a reduced incidence of breast cancers developing on tamoxifen in women who have a lower HR, but not in women with higher HR. There were too few BRCA1/2 mutations (4 cases) to be able to determine the efficacy of tamoxifen by BRCA status. Immunohistochemical analysis showed a significantly lower frequency of median ER (p=0.03) in the cancers developing in tamoxifen-treated patients. These results suggest that tamoxifen is less likely to be effective at reducing breast cancers which are ER negative and also in some individuals at higher HR.