The question of whether to offer adjuvant chemotherapy to patients with early-stage breast cancer continues to challenge clinicians on a daily basis. Many patients with node-negative or low-risk breast cancer could be spared the trauma of receiving chemotherapy, but more reliable prognostic markers are still needed to aid our therapy decision making. Gene expression profiling is a welcome product of the 'omic' era and several multi-gene expression panels ('signatures') have now been developed that show promise in predicting which breast cancer patients are likely to develop metastatic disease if adjuvant chemotherapy is not administered. The value of gene expression profiling as a prognostic tool in clinical practice is currently being appraised more fully in two large, prospective, randomised studies--TAILORx and MINDACT. These studies should provide level I evidence of the prognostic power of gene expression profiling and will hopefully allow us to one day quantify risk of progression in the individual patient and tailor treatment accordingly. Genetic profiling of circulating tumour cells and micrometastases should further enhance our understanding of breast cancer biology and allow us to personalise therapy based on functional maps of critical tumour pathways. Close collaboration between clinicians and scientists will be essential to achieve this goal.