The monoclonal antibody Ki67 recognizes an antigen expressed in all phases of the cell cycle except Go. It has been used in 141 biopsies from 138 patients with non-Hodgkin's lymphoma to identify proliferating cells in histological sections. A Ki67 index (the number of Ki67 positive tumour cells divided by the sum of Ki67 positive and negative tumour cells) has been derived by counting 1000 cells in each case. A correction for the presence of non-tumour cells has been incorporated by counting non-tumour cells in serial sections stained with a panel of other antibodies. A very strong correlation between a low Ki67 index (less than 20 per cent) and low grade histology and a high Ki67 index (greater than 20 per cent) and high grade histology was found (Chi2 = 98.0). Ninety-one patients could be analysed for survival and those with low grade lymphoma (n = 38) who had a relatively high Ki67 index (greater than 5 per cent) had a worse survival than those with an index of less than 5 per cent (p less than 0.05). In contrast, there was a trend for those patients with high grade disease with a very high Ki67 index (greater than 80 per cent) to have a better survival than those with a lower index (less than 80 per cent). The patients with high grade disease who achieved complete remission or good partial remission and had a Ki67 index of less than 80 per cent were more likely to relapse than those with an index of greater than 80 per cent (p less than 0.04). These findings could not be explained by the effect of other prognostic factors such as age, stage, or serum albumin. While the use of Ki67 immunostaining has potential drawbacks, it appears to be a simple and reproducible method of determining a tumour proliferative index which provides relevant clinical data.