Population pharmacokinetic-dynamic analysis was prospectively integrated in a broad phase II program of lurtotecan (GI147211), a novel camptothecin derived topoisomerase I inhibitor, to determine the population pharmacokinetic profile in a larger population, to estimate individual pharmacokinetic parameters and to investigate relationships with clinical outcome. A sparse sampling method was applied during course one, which involved two sampling time-points. A Bayesian algorithm was used to estimate individual pharmacokinetic parameters, in particular total plasma clearance (CL) and volume of distribution. In total, samples were collected of 109 (63%) of 173 patients. Pharmacokinetic-dynamic evaluation could be carried out successfully in 85 (78%) of the sampled patients. CL of lurtotecan showed substantial variability (mean 87 +/- 28 L/h) and was of the same magnitude as in the phase I studies where full pharmacokinetic curves were used. Residual variability in the population estimate of CL was 9.9%. No significant relationships were observed between exposure parameters and toxicity nor likelihood of tumor response, however the latter relationship may well have been obscured by the heterogeneity of the studied population. Prospective implementation of large scale population pharmacokinetic-dynamic analysis is feasible and important to establish whether interpatient variability in drug exposure is a major determinant of toxicity or activity.