A total of 161 previously untreated patients with FIGO stage III or IV epithelial ovarian cancer were randomised after surgery to receive six courses of either carboplatin 400 mg m-2 alone (Arm A) or carboplatin 300 mg m-2 with chlorambucil 10 mg day-1 for 7 days (Arm B). The median progression free survival (PFS) was similar: arm A: 45 weeks; arm B: 61 weeks (P = 0.830). Multivariate Cox regression analysis showed that the extent of residual disease and performance status were the most important prognostic factors for PFS. Fifty-two per cent of patients received dose escalations based on nadir blood counts, and 89% of all dose adjustments were made according to protocol. Failure to achieve a significant degree of leucopenia was associated with worse progression free survival (P less than 0.001). A total of 29.4% of patients fall into this category. The median survival was similar in both arms, i.e. 75 weeks. It is unlikely that there is any major clinical advantage to adding chlorambucil to single agent carboplatin for the management of advanced ovarian cancer, but whether used in combination or a single agent, the dose of carboplatin should be sufficient to cause at least grade I leucopenia. This may best be achieved by determining the initial dose based on renal function, and then adjusting subsequent doses according to nadir blood counts.