Estimation of S-phase fraction in breast carcinomas with single parameter flow cytometry may include errors due to dilution of cancer cells by host cells. Use of tissue specific markers may to some extent correct for the effect of dilution. S-phase fraction was estimated by flow cytometry with and without immunoselection in 80 DNA-euploid breast carcinomas in stage I-II. The tumor tissue was mechanically disintegrated and fixed in ethanol. A primary antibody, specific for cytokeratins 8 and 18, was added before incubation with the secondary FITC-conjugated antibody. S-phase fraction was calculated for both the gated (cytokeratin-positive) and the ungated cell population. An increasing proportion of tetraploid cells compared to diploid cells was found when the immunoselection method was used. The gated population tended to have a higher S-phase fraction than the ungated population. In univariate analysis S-phase fraction estimated from both ungated and gated cell populations yielded significant information for predicting recurrence when stratified for tumor size and nodal status. In bivariate analysis S-phase fraction of the gated population contributed prognostic information in addition to S-phase fraction of the ungated population when both variables were included in the analysis. Our conclusion is that S-phase fraction calculated from cytokeratin-positive cells provides prognostic information in addition to ungated S-phase values in DNA euploid breast carcinomas.