The analysis of thymidine labelling index (TLI) in relation to clinico-pathological variables and survival was carried out in 111 ovarian cancer patients. The significance of TLI in predicting response to aggressive first line chemotherapy regimens was examined. The overall median TLI value of 1.8% was used as a cut-off to discriminate slowly from highly proliferating cancers. 94 patients entered into two consecutive randomised trials, and were treated with six courses of cisplatin-based chemotherapy with or without doxorubicin. A significantly higher objective response of 60% was reported in the subset of patients with TLI > 1.8% as compared to 35% in patients with TLI < or = 1.8% (P = 0.03). In addition, patients achieving complete response had tumours with median TLI of 3.8% as compared to 2.4% for partial responders, 1.5% for patients with stable disease and 1.7% for those with progressive disease. A significant increase in tumour kinetics was observed in advanced cancers (P = 0.001), more undifferentiated tumours (P = 0.02) and postsurgical residual disease greater than 2 cm (P = 0.04). In univariate analysis, TLI failed to influence significantly clinical outcome: 26 versus 32 months median survival time for patients with high and low tumour TLI, respectively. In the Cox's regression model, the only independent prognostic variables were performance status and amount of residual disease after primary surgery (P = 0.000).