Many research projects in cell biology now use flow cytometry for analysis or for isolation of specific cell types. In such studies, cell viability is obviously a crucial issue. However, many studies appear to rely upon light-scattering characteristics to identify and gate out non-viable cells, despite the fact that reliable identification of such cells can only be achieved through staining with impermeable fluorescent nuclear dyes such as propidium iodide or 7-amino actinomycin. In this paper we apply mathematical analysis to the theoretical problem of quantifying cell sub-populations labeled with two or more fluorescent markers, comparing situations in which dead cells have been identified with those in which cell viability has not been assessed. We demonstrate that in all cases in which dead cells are present within the population, percentages of live sub-populations in different subsets are mis-estimated. In cases where the pattern of marker expression differs greatly between live and dead cells, or where the proportion of dead cells is high, this mis-estimation will be aggravated; the subsets pattern will therefore be biased in a population selected only on the basis of light-scatter behavior. The importance of accurately detecting and gating out dead cells is illustrated by an experimental example accompanying the mathematical analysis. To conclude, identification of dead cells by means of viability stains should be an absolute routine in practical flow cytometry, so as to avoid mis-estimation in sorting or analysis.