Drawing upon the comprehensive population-based Northern Alberta Breast Cancer Registry containing 704 patients with histologically negative axillary lymph nodes who have been followed for 5-16 years, we have undertaken a retrospective case-control study to evaluate the utility of genomic amplification of specific protooncogenes [c-erbB-2 (nee HER-2/neu), c-erbA, c-myc, int-2, and hst-1] as predictive indicators of clinical outcome in node-negative disease. To this end, 115 women with node-negative breast cancer who had recurred at any time up to 16 years posttreatment (cases) were matched pairwise for appropriate clinicopathological variables (size of primary tumor, menopausal state, estrogen receptor status, anniversary year of treatment, and patient age) with a second group of 115 women (controls) selected from a cohort of 502 node-negative patients who had not relapsed during long-term follow-up. Tumor DNA extracted from archival formalin-fixed, paraffin-embedded tissue blocks were analyzed for protooncogene copy number by slot-blot hybridization. Taking a gene copy number of 3 as the cutoff, 27 of the 230 tumor samples examined contained from 3- to 22-fold elevation in c-erbB-2 genomic equivalents. Twenty-one of the 27 tumors amplified for c-erbB-2 were derived from cases and 6 from controls, signifying that 18% of the node-negative patients who had relapsed harbored excessive copies of the protooncogene in their malignant tissue compared to only 5% for the patients who had remained in remission. Accordingly, the occurrence of amplification of c-erbB-2 proved to be a statistically significant predictor of poor prognosis, especially disease-free interval (P = 0.006). Moreover, this genetic alteration appeared to be independent of and to have greater predictive power than most commonly used prognostic factors. Our findings also indicated that as a clinical test, measurement of c-erbB-2 amplification suffers from low sensitivity; however, when greater than 6 gene copies are present, the test has a positive predictive value for recurrence of 70%. Concurrent analysis of tumor DNA blots with probes for the other four protooncogenes examined revealed that their amplification, which others have reported to arise often, especially in node-positive disease, was seldom found even in our high-risk case group (2-3%). In short, our data strongly suggest that amplification of c-erbB-2 may contribute to the pathogenesis of some forms of node-negative breast cancer and thus may serve as a useful genetic marker to identify a subset of high-risk patients.