The clinical usefulness of histologic grading in follicular lymphoma (FL) is controversial and is further compromised by the subjective nature and poor reproducibility of most systems in current use. Therefore, we decided to objectively evaluate the importance of cellular proliferation in FL, along with the current grading systems. We studied 106 patients with FL who were uniformly staged and aggressively treated. A proliferative index (PI) was determined quantitatively using an automated image analyzer and a new Ki-67 antibody that stains archival paraffin tissues. The cases were also subclassified according to the Berard, Rappaport, Luke-Collins, and Jaffe methods, and survival analysis was performed. Patients with a low PI (< 40%) had a significantly longer overall survival (OS) than those with a high PI (> or = 40%), but the PI did not predict failure-free survival (FFS). The mean PI correlated well with the subgroups in each of the various classifications. All four of the classification methods were predictive of OS, but only the Berard method appeared to predict FFS and suggest that a proportion of patients with FL may be curable. In multivariate analysis, histologic classification was the only independent predictor of OS (Berard method: relative risk, 3.1) and the International Prognostic Index was the only independent predictor of FFS (relative risk, 2.3). We conclude that the Berard method for grading of FL is clinically useful and, along with the International Prognostic Index, should be included in future clinical studies of FL. The measurement of cellular proliferation does not appear to add additional useful information in FL.