By retrospective analysis of 13,000 blood samples obtained from 248 patients receiving FK 506 therapy, we compared the suitability of plasma with that of whole blood as the matrix for therapeutic drug monitoring of FK 506. The plasma concentrations did not correlate with the concentrations in whole blood (r = 0.56). In contrast to plasma samples (analyzed by enzyme immunoassay), FK 506 was detectable in all whole-blood samples (analyzed by enzyme immunoassay/microparticle enzyme immunoassay). The inter- and intraindividual variations of FK 506 measurements were greater in plasma than in whole blood. Moreover, plasma concentrations correlated only poorly with clinical events. There was a tendency to greater plasma concentrations being measured during episodes of toxicity, but no clear difference was evident between stable course and rejection. In whole-blood specimens, a correlation between reduced or increased FK 506 concentrations and rejection or toxicity, respectively, was observed. The discriminatory power of whole-blood values was greater for the differentiation between toxicity and stable course than between rejection and stable course. We therefore recommend whole blood rather than plasma as the matrix for therapeutic monitoring of FK 506 concentrations.