5,6-Dimethylxanthenone-4-acetic acid (DMXAA), a low-molecular-weight biological response modifier scheduled for clinical evaluation, induced synthesis of tumour necrosis factor-alpha (TNF-alpha) in serum of mice, with maximal activity being observed at 2-3 h after administration. At a dose of 27.5 mg/kg, DMXAA induced similar TNF-alpha concentrations as did flavone-8-acetic acid given at its maximum tolerated dose (MTD; 330 mg/kg), whereas 8-methylxanthenone-4-acetic acid, which has no antitumour activity, did not induce serum TNF-alpha at its MTD (440 mg/kg). The dependence of schedule on TNF-alpha induction was studied by giving DMXAA to mice in two doses of 27.5 mg/kg each separated by different intervals. An interval of 0 (i.e. 55 mg/kg given in a single dose) produced a TNF-alpha concentration 9-fold that produced by a single dose of 27.5 mg/kg. This dose, although higher than the MTD of 30 mg/kg, did not affect the health of mice at the time of assay (3 h). An interval of 1 day produced very low levels of serum TNF-alpha after the second injection. An interval of 3 days produced high levels of serum TNF-alpha after the second injection (9-fold that detected in mice receiving 27.5 mg/kg in a single dose) but no long-term toxicity, whereas an interval of 7 days produced an intermediate response. Thus, the first dose can either potentiate or suppress the TNF-alpha response to a second dose. Mice with advanced subcutaneous colon 38 tumours were treated either with a single dose of DMXAA (27.5 mg/kg) or with a divided dose (two doses of 27.5 mg/kg given 3 days apart). Both the cure rate and the tumour-growth delay were enhanced by the divided-dose schedule. The results are relevant to the design of clinical administration schedules of DMXAA and emphasise the importance of TNF-alpha induction in the antitumour response.