It is now accepted that c-erbB2-gene amplification is correlated with poor clinical outcome for patients, mainly when axillary nodes are invaded. We have confirmed this result by multivariate analysis in 178 patients with non-inflammatory breast cancer followed up for a mean period of 6.8 years (SD, 1.6 years). In addition, we have shown that c-erbB2 amplification, found in 30 (17%) specimens, was associated with a high risk of multiple metastases developing simultaneously. In contrast, for the 67 patients with inflammatory breast carcinoma, the most aggressive type of breast carcinoma, the c-erbB2 amplification detected in 24 (36%) specimens was not found to be associated with a higher risk of death, suggesting that the c-erbB2 gene plays a different role in the progression of these 2 types of breast cancer. Furthermore, our data stress the importance of the methodological approach used to determine gene amplification. Although Southern blot hybridization is a tumour- and time-consuming method not easy to adopt in routine clinical practice, this method remains a reference quantitative method.