Context
The most important prognostic indicators to predict patient outcome in breast cancer are hormone receptor status, axillary lymph node status and histological grade; although these give a broad overview of outcome, they cannot predict for all patients. The recent development of cDNA array analysis provides the opportunity for a more refined approach based on molecular classification of individual tumours (see Additional information [1]). A previous study by the same authors identified molecular profiles in a cohort of nine benign and 72 malignant breast tumours (see Additional information [2]). The aim of this work was to use this technology to identify patients at increased risk of tumour recurrence.