The ability of a wide variety of anionic, cationic, and neutral drugs to bind in a reversible manner to plasma proteins has long been recognised. Non-steroidal anti-inflammatory drugs (NSAIDs) are distinguished as a class by the high degree to which they bind to plasma protein. Plasma protein binding properties are primary determinants of the pharmacokinetic properties of the NSAIDs. Theoretical relationships are reviewed in order to define quantitatively the impact of plasma protein binding on clearance, half-life, apparent volume of distribution, and the duration and intensity of pharmacological effect. The quantitative relationships governing competitive displacement binding interactions are also presented. Experimental methods for in vitro and in vivo determination of the degree of plasma protein binding are discussed. The more common in vitro methods are equilibrium dialysis and ultrafiltration. Methods for characterising the degree of plasma protein binding in vivo consist of either measuring the concentration of drug at equilibrium in an implanted semipermeable vessel or measuring the relative drug concentrations in two body spaces with different protein content. Emphasis is given to the comparative advantages and disadvantages of experimental application of the various in vitro and in vivo methods. Plasma protein binding is discussed as a determinant of the trans-synovial transport of NSAIDs. Trans-synovial transport of NSAIDs appears to be a diffusional process. Limited data in humans receiving ibuprofen, indomethacin, aspirin, carprofen, alclofenac, or diclofenac suggest that clearance of each of these NSAIDs from the synovium is slower than clearance from plasma. The clinical data relevant to the relationship between plasma NSAID concentration and various measures of anti-inflammatory effect are reviewed. A positive correlation between plasma NSAID concentration and anti-inflammatory effect has been observed in only one study on naproxen and one study on piroxicam. In several other studies, the lack of concentration-response correlations is generally attributed to the relatively subjective, quantitatively inexact methods used to assess anti-inflammatory effect and analgesia in arthritic patients, as well as the substantial interpatient variabilities in the fraction of unbound NSAID and the unbound plasma NSAID concentration. In view of the generally poor correlation between concentration and therapeutic response, routine therapeutic monitoring of total plasma NSAID concentration is not recommended as a means of titrating individual dosages to the desired effect in each patient.(ABSTRACT TRUNCATED AT 400 WORDS)