OBJECTIVE Paracetamol has a central action for both antipyresis and analgesia. Maximum temperature decrease and peak analgesia are reported at 1-2 h after peak plasma paracetamol concentration. We wished to determine the relationship between plasma and cerebrospinal fluid (CSF) pharmacokinetics in children. METHODS Concentration-time profiles in plasma and CSF after nasogastric paracetamol 40 mg kg(-1) were measured in nine children who had indwelling ventricular drains. Estimation of population pharmacokinetic parameters was made using both a standard two-stage population approach (MKMODEL) and a nonlinear mixed effect model (NONMEM). Results were standardized to a 70 kg person using an allometric power model. RESULTS Both approaches gave similar estimates. NONMEM parameter estimates were clearance 10.21 h(-1) (CV 47%), volume of distribution 67.11 (CV 58%) and absorption rate constant 0.77 h(-1) (CV 49%). Cerebrospinal fluid concentrations lagged behind those of plasma. The equilibration half time was 0.72 h (CV 117%). The CSF/plasma partition coefficient was 1.18 (CV 8%). CONCLUSIONS Higher concentrations in the CSF probably reflect the lower free water volume of plasma. The CSF equilibration half time suggests that CSF kinetics approximate more closely to the effect compartment than plasma, but further time is required for paracetamol to exert its effects. Effect site concentrations equilibrate slowly with plasma. Paracetamol should be given 1-2 h before anticipated pain or fever in children.