The thymidine analogue zidovudine is currently used for the treatment of HIV-infected patients, as the early development of the drug as an anticancer agent yielded modest results. A comprehensive preclinical analysis, however, showed that inhibitors of de novo thymidylate synthesis, including fluorouracil and methotrexate, enhanced the antiproliferative activity of zidovudine in cancer cells. Significant inhibition of tumour growth was obtained in mice bearing human colon cancer xenografts and given intraperitoneal zidovudine 300 to 600 mg/kg weekly in combination with methotrexate 87.5 mg/kg or intraperitoneal fluorouracil 85 mg/kg, and in pharmacokinetic studies high peak drug plasma concentrations (Cmax) of zidovudine were obtained, ranging from 610.3 to 1698.8 mumol/L. In order to exploit the therapeutic activity of zidovudine, phase I and II clinical studies were designed in combination with fluorouracil and the pharmacokinetic-pharmacodynamic profile of zidovudine was investigated. Clinical responses were obtained in patients treated intravenously with bolus fluorouracil 500 mg/m2, leucovorin and short (90 to 120 minutes) infusions of high dose zidovudine (up to 10 g/m2), generating drug Cmax similar to those obtained in preclinical models. However, in chemotherapy-pretreated patients receiving high dose zidovudine by the oral route (1 to 9 g/m2/day) or 48-hourly continuous intravenous infusion (2 to 20 g/m2/day) in combination with fluorouracil and leucovorin, treatment failures were observed despite high systemic exposure, described as the area under the plasma concentration-time curve and the occurrence of DNA strand breaks in peripheral blood mononucleated cells, the biological expression of zidovudine activity. In conclusion, preclinical and clinical evidence suggest that the schedule of administration of zidovudine is a requisite for the expression of its activity, indicating the importance of concentration-monitored trials to optimise chemotherapy dose administration in patients. The likelihood of tumour response appears to be related to the achievement of high peak plasma concentrations of zidovudine, and constant infusions appear less likely to produce clinical results.