The pharmacokinetics of one of the most widely used non-steroidal antiinflammatory drugs, naproxen, were studied in 28 healthy human volunteers at the two most commonly used dose levels, viz., 250 mg and 500 mg, in a cross-over design. The plasma levels of naproxen were analysed by a modified high-pressure liquid chromatography method. The plasma concentrations at higher doses were not proportional to dose, indicating a non-linearity in the pharmacokinetics at the dose levels studied; this finding is new since earlier studies had studied only higher doses and assumed that at lower doses the pharmacokinetics would be linear. There was, however, no significant difference in the elimination half-life (rate constant), time to reach peak concentration (Cmax), mean residence time (MRT), or area under first moment curve (AUMC). The clearance and distribution volume of naproxen were substantially increased at higher dose resulting in statistically lower proportional concentration and the total area under the curve (AUC). These observations are explained on the basis of a change in the plasma protein binding resulting in more free naproxen available for quicker clearance and wider penetration into tissues. These findings have several important clinical implications for the long-term use of naproxen as an antiarthritic drug. It is proposed that the clinical efficacy of naproxen can be increased and side-effects reduced by giving it in small divided doses instead of large doses.