Dihydropyrimidine dehydrogenase (DPD) is the initial enzyme of 5-fluorouracil (5-FU) catabolism. The clinical importance of DPD has recently been demonstrated with the identification of severe and/or lethal 5-FU-related toxicity in patients with suspected or proven DPD deficiency revealed in lymphocytes. We conducted a prospective study on 185 cancer patients in order to evaluate the incidence of complete or partial DPD deficiency. The population comprised 152 men (mean age 62.1) and 33 women (mean age 59.2). Sixty eight were head and neck patients treated by a 5-day continuous infusion of 5-FU (starting dose 1 g/m2/day, with dose adaptation at mid-cycle) for which DPD activity was measured 2-3 days before 5-FU administration (94 cycles analyzed). DPD activity was measured by a radioenzymatic assay using 14C-5-FU. DPD activity showed a unimodal distribution, which globally fits a Gaussian distribution. Mean and median DPD activity were 0.222 and 0.211 nmol/min/mg prot respectively (range 0.065-0.559). No total DPD deficiency was found. Neither liver function nor age influenced DPD activity, but DPD activity was on average 15% lower in women than in men (0.194 and 0.228 nmol/min/mg prot respectively, p = 0.03). In patients treated with 5-FU, the risk of developing side effects was not linked to pretreatment DPD activity. 5-FU-related toxicity was linked to FU systemic exposure. The correlation between DPD activity and FU clearance was weak (n = 90, r = 0.31, p = 0.002). As a corollary, DPD activity in patients requiring a dose reduction was not significantly different from DPD activity in patients who did not require dose modification. From the present study it appears that total DPD deficiency is a very rare event. Although pre-treatment DPD activity cannot be a useful indicator for improving 5-FU dose adaptation strategy, the identification of partial DPD deficiency (< 0.100 nmol/min/mg prot, 3% of the population) could lead to starting the treatment with a markedly reduce 5-FU dose.