We report a novel approach to gene expression profiling using the Nottingham Prognostic Index (NPI) to stratify 26 patients with invasive breast carcinoma. As an aggregate index of parameters reflecting metastatic potential, growth rate, and genetic instability the NPI has distinct advantages over other clinicopathologic features used to segregate breast cancer patients. As a continuous variable it offers a responsive and sensitive means of modeling a continuum of clinical aggressiveness. Using RNA extracted from 26 tumors and cDNA microarrays with 23 343 unique genetic elements, 84 genes and expressed sequence tags were identified whose expression patterns correlated with NPI. Differential expression by immunohistochemistry (IHC) was also observed for two of three genes evaluated by this method. Correlation was determined by the Spearman rank correlation method with null distribution analysis. Among the 84 genetic elements were seven previously implicated in neoplastic progression (including the two demonstrating differential expression by IHC), 11 without specific cancer association but localized to chromosomal sites whose loss or gain has been identified in cytogenetic studies of breast carcinoma, and 73 not previously associated with breast carcinoma. Collectively, the expression patterns of these 84 elements have potential to distinguish high and low NPI patient samples. These data add support to the assertion that prognostic groups of breast carcinoma are reflected in distinguishable expression profiles of a limited set of genes.