The clinicopathological features currently used in breast cancer prognosis often fail to characterize the clinical heterogeneity of the disease accurately. Our study is aimed to investigate the predictive value of DNA flow cytometry in breast cancer. Previously untreated breast carcinoma samples (584) were snap frozen for flow-cytometry. Tumors were classified into three DNA index (DI) categories: i) tumors showing a DI =0.96-1.15 (diploid and near-diploid); ii) tumors with a DI >or=1.16 (hyperdiploid, tetraploid, multiploid and/or those with more than one diploid population); and iii) tumors with a DI <or=0.95 (hypoploid). The 5- and 10-year cumulative survival rates +/- SE for Group I (n=191) were 98+/-1% and 98+/-1%. For Group II (n=361) these rates were 77+/-2% at 5 years and 63+/-5% at 10 years. In Group III (n=32) the rate at 5 years was 23+/-8%, with no patients alive at 10 years (p<0.0001). In univariate analysis, tumor size, node status, grade, karyometry, S-phase fraction, MIB-1 index, and estrogen receptors retained prognostic significance; in multivariate analysis, only DI <or=0.95 (hypoploid) was retained as an independent prognostic factor for overall survival. Our data strongly support that DNA hypoploid has a strong, independent prognostic value for predicting the short-term clinical outcome of breast carcinoma patients.