The relationship between immunoreactivity for cell proliferation markers (Ki67 and PCNA) and the growth fraction as determined by the fraction of labelled mitoses method was assessed in xenograft tumours grown from the LoVo cell line in nude mice. FLM curves were constructed by injecting tritiated thymidine and then preparing autoradiographs from the tumours. From this data an estimate of growth fraction and cell cycle time were made. Using frozen material from the same tumours, the Ki67 index was determined by immunostaining. PCNA staining was determined in the fixed material which had been used for the autoradiographs. The results show that Ki67 staining follows the same trend as the FLM-determined growth fraction as the tumour increases in size and the rate of growth decreases. However the Ki67 index does produce a consistent overestimate of the growth fraction in this in-vivo system, as compared to observations in cell culture. PCNA staining showed virtually 100 per cent positive staining in all the tumours, which is likely to reflect the long half-life of the antigen, compared to the very fast cell-cycle time of the xenograft tumours. These results show that staining with proliferation markers is not a precise determinant of growth fraction. Ki67 staining is a method that can be usefully applied as an operational marker of cell proliferation, but should not be used uncritically. Further caution is necessary in the use of PCNA. The findings also demonstrate the need to use a range of methods when assessing a new proliferation marker.