The recent discovery of cyclooxygenase-2 (COX-2), an isoenzyme associated mainly with inflammation created the need to reevaluate cyclooxygenase inhibitors with reliable screening methods. In the present study we standardized a technique to determine the IC50S of cyclooxygenase inhibitors on recombinant human COX-1 and COX-2 expressed in mammalian cells and used it to study the compounds tenoxicam, aspirin and indomethacin. The IC50S of aspirin, indomethacin and tenoxicam for human COX-1 were 0.41 +/- 0.07 microgram/ml, 0.008 +/- 0.003 microgram/ml, and 7.94 +/- 3.28 micrograms/ml, respectively, and for human COX-20.64 +/- 0.16 microgram/ml, 0.09 +/- 0.05 microgram/ml, and 10.61 +/- 1.50 micrograms/ml, for aspirin, indomethacin, and tenoxicam. Tenoxicam had the lowest IC50hCOX-2/IC50hCOX-1 ratio (1.34), followed by aspirin (1.53) and indomethacin (10.82). The system described in the present study provides a simple and efficient way to determine the specificity of NSAID inhibition for each of the human cyclooxygenase isoenzymes separately.