OBJECTIVE Available data on colistin pharmacokinetics in patients undergoing continuous renal replacement therapy (CRRT) are limited. Our aim was to study colistin pharmacokinetics in critically ill patients treated with colistin methane sulphonate for Gram-negative sepsis and undergoing continuous venovenous haemodiafiltration for acute renal failure. METHODS Three patients were studied. The colistin methane sulphonate dose administered was at the discretion of the attending physician and was in all cases lower than that recommended for individuals with intact renal function. Colistin methane sulphonate was administered intravenously over 30 min, and blood samples were collected from each patient pre- and post-filter for the HPLC determination of colistin levels in serum before infusion, at 10, 60, 120, 240, 360, 480 and 600 min from the end of infusion, and immediately before the next dose. Concurrently, spot samples of effluent from the haemofilter were also collected and analysed. Both colistin total extracorporeal clearance and clearance in the effluent were calculated. RESULTS Extracorporeal clearance resulted in substantial removal of colistin (43%-59% of total colistin clearance). Total colistin clearance was found to be reduced (varying between 3.3 and 4.5 L/h), compared with patients with normal renal function. Colistin methane sulphonate dosage resulted in clearly suboptimal colistin steady-state concentrations. CONCLUSIONS In spite of substantial extracorporeal clearance, total colistin clearance was reduced, compared with patients with normal renal function. Colistin adsorption by the haemofilter contributed to its extracorporeal clearance to a large extent. Studies on other patients receiving colistin methane sulphonate and undergoing CRRT are required before more appropriate dosage regimens can be recommended.