We previously have shown that pharmacokinetic monitoring of 5-fluorouracil (5-FU) could significantly improve the 5-FU therapeutic index when given in continuous venous infusion (Br J Cancer 59, 287-290, 1989). However, the more rational approach would be to find individual biological factors which could predict 5-FU clearance. Dihydropyrimidine dehydrogenase (DPD) is the initial enzyme in the catabolism of 5-FU. DPD activity was measured in 57 head and neck cancer patients receiving CDDP (100 mg/m2, day 1) plus 5-FU (1 g/m2/day x 5, day 2-day 6). DPD activity was measured in lymphocytes using a radioenzymatic assay (2.5 mM MgCl2, 250 microM NADPH, 20 microM 14C-5-FU) with separation of 14C-5-FU from 14C-5-FUH2 by HPLC coupled with a radiodetector. The average DPD activity measured in lymphocytes was 0.186 +/- 0.068 nmol/min/mg protein (range 0.058-0.357) and the average 5-FU clearance (Cl) was 2,523 +/- 684 ml/min/m2 (range 1,052-4,029). A significant linear correlation was demonstrated between DPD activity and 5-FU clearance (Cl = 1,099 + 7,580 DPD, r2 = 0.613, P < 0.0001). In patients evaluated for more than one cycle (n = 18), variations in 5-FU clearance were associated with corresponding variations in DPD activity. The individual determination of DPD activity measured in lymphocytes could be useful for identifying patients at risk for altered 5-FU pharmacokinetics and could be used to adjust the optimal 5-FU dose for each patient before starting the treatment.