Imatinib pharmacokinetics (PK) may be affected by a number of factors that are related to the disease being treated and to the response of that disease to imatinib. Patients in the phase I and phase II trials conducted by the EORTC in patients with gastrointestinal stromal tumours (GISTs) and other sarcomas had detailed blood sampling for imatinib PK on day 1 and on day 29. Patients with GISTs also had repeat sampling, using a limited sampling strategy, after approximately 12 months on therapy. This population PK study was carried out to examine what covariates affected imatinib PK in GIST patients and what PK changes occurred over time. In the model producing the best fit, low clearance (CL) correlated with low body weight and high granulocyte count, whereas low haemoglobin correlated with low volume of distribution. For a patient with 77% of the median body weight or with 1.87 times the median granulocyte count, the apparent CL is 6.53 l/h, about 70% of the typical apparent CL of 9.33 l/h; for a patient of 84% of the typical haemoglobin level, the volume of distribution is about 70%. Total white blood cell count correlated closely with granulocyte count and there was a moderate correlation between haemoglobin and albumin (r = 0.454). There was a trend towards increased imatinib clearance after chronic exposure over 12 months. The typical apparent CL increased 33% from day 1. Nevertheless, the approximate 95% confidence interval of the increase of the typical apparent CL was 33 +/- 34.6%, which contains zero. It is not yet clear whether this is a significant factor in the amelioration of imatinib toxicity that occurs with time or is related to disease control, and further work is required to confirm this observation.