Gene amplification, an important mechanism of oncogene activation in breast cancer, can have both prognostic and therapeutic implications. In this work, an attempt is made to identify amplified genes that can be used to improve prognostication in breast cancer. A series of 52 node-negative tumours was screened for genomic gains at 57 loci by array-CGH. A subset of these genes was identified that could divide the series into two divergent outcome groups of either long-term survivors or early disease-related deaths (p = 0.01) using a combination of k-means clustering and statistical analysis. The prognostic significance of amplification of four of the genes (EMS1, TOP2A, CCNE1, and ERBB2) was then evaluated, using fluorescent in situ hybridization on a tissue microarray, in a second larger 'validation' series of 232 tumours with a median follow-up of 4.8 years. Adverse disease-related outcome was associated with amplification of TOP2A (p = 0.004); ERBB2 (p = 0.002); and with the combined amplification of TOP2A, ERBB2, and EMS1 (p = 0.01). EMS1 amplification was more common (26% of cases) than previously reported but, in isolation, had no prognostic significance. Amplification of CCNE1, seen in only 6% of cases, had no prognostic role. These results indicate that the complementary use of array-CGH and tissue microarrays has the potential to help in the identification and validation of molecular markers that can be used to classify breast cancers into different prognostic groups.