Paclitaxel is efficacious against many human cancers. Because it blocks cells at the radiosensitive G2-M interface, paclitaxel has been investigated as a radiosensitiser. The results have been equivocal and somewhat contradictory. It is impossible to obtain proper pharmacokinetic calculations, aimed at obtaining maximum cytotoxicity and/or radiosensitisation, without knowing (i) how long the drug must be in contact with the cells, (ii) how long the effect lasts after the drug is removed from the cellular environment, (iii) whether the drug acts as a radiosensitiser even when, like cis-platinum, it is added after the radiation and (iv) what the minimum quantity of drug in the cellular environment is required for both chemotoxicity and radiosensitisation. The present work addresses the above questions. Two radioresistant cell lines of human origin were used, A375 melanoma and S549 lung carcinoma, in a clonogenic assay where only colonies with 50 or more cells were counted. For the irradiation, 6 MV X-rays were used. Any G2-M block was quantified by cell cycle kinetics analysis. From the results, a simulation of pharmacokinetics was conducted to calculate the schedule of administration of paclitaxel most likely to achieve and maintain significant chemotoxocity and radiosensitisation. The minimum concentration of paclitaxel for measurable cytotoxicity was 3 nM for both cell lines, but the drug was more toxic to the A549 cells. The minimum concentration for measurable radiosensitisation was 3 nM for A375 and approximately 0.1 nM for A549, but whereas above 3 nM the radiosensitivity increased in A375, it decreased above 1 nM for A549. A minimum of 18 h incubation with the drug was necessary for measurable effects and the radiosensitising effects were lost soon after its removal. There was no radiosensitisation if paclitaxel was added after the radiation, and, at the minimum effective concentrations, it caused only a minor and transient G2-M block. The pharmacokinetic calculations predict that 15 mg/m2 paclitaxel given as a 1 h infusion 5 days/week for 3 weeks during the radiotherapy should achieve both cytotoxicity and radiosensitisation. The mechanism of radiosensitisation by paclitaxel at the concentrations suggested by our results does not appear to be via a G2-M block and is probably concentration dependent. The results imply that low-dose, daily infusions of paclitaxel for as long as possible during a course of radiotherapy are more likely to result in radiosensitisation and prolonged cytotoxicity than high-dose infusions given once a week.