The function of the tumor suppressor RE1 silencing transcription factor (REST) is lost in colon and small cell lung cancers and is known to induce anchorage-independent growth in human mammary epithelial cells. However, nothing is currently known about the role of this tumor suppressor in breast cancer. Here, we test the hypothesis that loss of REST function plays a role in breast cancer. To assay breast tumors for REST function, we developed a 24-gene signature composed of direct targets of the transcriptional repressor. Using the 24- gene signature, we identified a previously undefined RESTless breast tumor subtype. Using gene set enrichment analysis, we confirmed the aberrant expression of REST target genes in the REST-less tumors, including neuronal gene targets of REST that are normally not expressed outside the nervous system. Examination of REST mRNA identified a truncated splice variant of REST present in the REST-less tumor population, but not other tumors. Histological analysis of 182 outcome-associated breast tumor tissues also identified a subpopulation of tumors that lack full-length, functional REST and over-express the neuroendocrine marker and REST target gene Chromogranin A. Importantly, patients whose tumors were found to be REST-less using either the 24-gene signature or histology had significantly poorer prognosis and were more than twice as likely to undergo disease recurrence within the first 3 years after diagnosis. We show here that REST function is lost in breast cancer, at least in part via an alternative splicing mechanism. Patients with REST-less breast cancer undergo significantly more early disease recurrence than those with fully functional REST, regardless of estrogen receptor or HER2 status. Importantly, REST status may serve as a predictor of poor prognosis, helping to untangle the heterogeneity inherent in disease course and response to treatment. Additionally, the alternative splicing observed in REST-less breast cancer is an attractive therapeutic target.