A novel enzyme immunoassay for plasma thrombospondin (TSP) based on commercially available monoclonal antibodies was established. The following conditions for correct collection and preservation of blood samples were required: venipuncture directly into a vacutainer containing citrate, theophylline, adenosine and dipyridamole, storage on ice, and separation of plasma within 30 minutes. Thereafter, the plasma TSP concentration remained constant at room temperature and after five times of freezing and thawing. Both inter- and intraassay variation coefficients were 5%. The lower detection limit was 20 microg/L. Median TSP concentration among 40 healthy blood donors was 43 microg/L, slightly lower than previously published. The assay is valid, reliable, and has certain advantages compared with previously published methods. TSP and beta-thromboglobulin (BTG) were then compared as platelet activation and biocompatibility markers in vivo: 23 patients undergoing cardiopulmonary bypass (CPB); and in vitro: effect of coating polyvinyl chloride with heparin. The kinetic patterns of TSP and BTG were markedly different in vivo but virtually identical in vitro, explained by different in vivo clearance mechanisms during CPB. We conclude that BTG is superior to TSP for evaluation of platelet activation during in vivo CPB, whereas TSP and BTG are virtually identical as markers in vitro.