The ERBB2 proto-oncogene was studied in 539 invasive primary breast tumors and was found amplified (2- greater than 30 copies) in 19%. Amplification was correlated to most known risk factors, including; large tumor size, lymph node positivity and many tumor involved nodes, advanced stage, low patient age (less than 40 years), non-diploidy and hypertetraploidy, and most significantly (P less than 0.00001) to the absence of steroid receptors and to a high rate of proliferation (flow cytometric determined S phase fraction). ERBB2 amplification was strongly associated (P less than 0.0001) with early recurrence and death in breast cancer among node-positive patients. This connection did not, however, remain in multivariate analyses. No correlations to clinical outcome were seen among node-negative patients. Similarly, non-diploid, but not diploid, amplified tumors were particularly aggressive. Furthermore, ERBB2 amplification was associated with a high rate of proliferation and poor prognosis in steroid receptor positive, but not receptor negative tumors. In progesterone receptor positive breast cancer, amplification was an independent and with node status equally powerful (P less than 0.0001) predictor of poor survival. It is concluded that ERBB2 activity is related to an increased tumor growth rate but not directly to metastasizing ability. Its clinical relevance as a prognostic factor may be in selecting a high risk subgroup of breast cancer, in general considered as being of good prognosis.