Nuclei, isolated from paraffin-embedded tissue, were stained with propidium iodide (PI) and found suitable for DNA analysis by flow cytometry (FCM). DNA-derived fluorescence intensity, however, was always decreased and had a much higher intersample variability as compared to results obtained with fresh material. Using chicken red blood cells (CRBC) as a model system, we found the lower fluorescence intensity to be due to the formalin fixation step in tissue processing. The intersample variability was found to be at least partly caused by variations in the duration of fixation. Overnight trypsinization improved the fluorescence intensity but did not reduce the intersample variability. Under all conditions tested PI binding to CRBC appeared to be saturable. Since fresh diploid or red blood cells could not be used to standardize DNA histograms, an alternative approach was developed in which nuclei from paraffinembedded normal and tumor tissue of the same specimen were mixed. With this method DNA indices (DI) of 24 colorectal cancers were found to be closely correlated (r = 0.9877, P < 0.001) with DI obtained with fresh tumor tissue from the same patients. The correlation of the percentages of S-phase nuclei between paraffin-extracted and fresh samples (r = 0.5875, P < 0.05) was as high as could be expected, taking sampling differences into account. This method is an important tool for the retrospective analysis of FCM-derived DNA parameters in relation to diagnosis and prognosis of neoplasms.