A number of different factors can profoundly influence the quantification of immunostained cells. Given the characteristics of immunohistological detection systems with non-linearity of signal and antigen concentration, we investigated the relationship of signal (number of stained cells) to the dilution of antibody employed. Three antibodies were studied which have been advocated as being effective in fixed material as markers of cell proliferation: PC10 (anti-proliferating cell nuclear antigen (PCNA)), Ki-S1 and MIB1 (a novel anti-Ki-67). Serial sections of tonsil were immunostained with a range of antibody dilutions using a fixed detection system and the number of stained cells quantified. Similar experiments were performed on tumour xenografts with known growth fraction and, in vitro, on human diploid fibroblasts in logarithmic growth phase. With both PC10 and Ki-S1 the number of stained cells increased with decreasing antibody dilution with no plateau being identified. In contrast, MIB1 showed a clear plateau. Immunocytological data indicate that PCNA and Ki-S1 antigen are present at low (but detectable) levels in at least some non-cycling cells and thus an artificial 'cut-off' has to be employed in assessing the number of proliferating cells with these antibodies. The superiority of MIB1 probably reflects the rapidity of catabolism of the Ki-67 antigen at the end of M phase. Taken together, these data point to the importance of carefully considering fundamental immunochemical properties such as antibody concentration (as well as antibody affinity and sensitivity of detection system) when employing immunological markers of cell proliferation in quantitative procedures.