DNA-variables such as DNA-ploidy, DNA-index and %S-phase cells have been proven to have prognostic value for breast cancer patients. These variables can be obtained by interpreting DNA-histograms by cell cycle analysis. Since there are a number of potential error sources, the aim of this study was to determine the intra- and inter-observer reproducibility of semi-automated cell cycle analysis with the emphasis on DNA ploidy and %S-phase assessments. The 149 DNA-histograms we used were randomly selected from the Multicentre Morphometric Mammary Carcinoma Project, a nationwide prospective study in The Netherlands on the reproducibility and prognostic power of quantitative assessments. These DNA-histograms were obtained by flow cytometry of fresh frozen breast cancer material. Cell cycle analysis was performed according to a strict protocol with the semi-automated computer program 'MultiCycle', using the background correction option; 68 histograms were analyzed in duplicate by the same observer, and 81 histograms were analyzed by two observers. Assessment of DNA-ploidy showed an intra-observer concordance of 99% (kappa-value 0.98) and an inter-observer concordance of 94% (kappa-value 0.91). The disagreement could be attributed to overlooking a DNA-tetraploid cell cycle in one case, some difficult histograms and varying opinions about small peaks between the observers in a few cases. Intra-observer %S-phase correlation coefficients varied between 0.72 for the %S-phase of the second aneuploid cell cycle and 0.99 for the %S-phase of the diploid cell cycle. Inter-observer correlation coefficients varied between 0.81 for the %S-phase of the second cell cycle and 0.95 for the %S-phase of the diploid cell cycle and the average %S-phase cells. As for DNA-index, intra- and inter-observer correlation coefficients were 0.97 and 0.94, respectively. In conclusion, intra- and inter-observer reproducibility of semi-automated cell cycle analysis of flow cytometric DNA-histograms from fresh breast cancer material using the Multi-Cycle program, following a strict protocol, is in principle high. The results of this study may help us to decide which of the different %S-phases provided by cell cycle analysis software should be used in daily practice.