An international 19-laboratory survey was organized to compare assays for 25-hydroxyvitamin D, 24,25-dihydroxyvitamin D, and 1,25-dihydroxyvitamin D in plasma. Each participant received two ethanolic standard solutions of each metabolite and eight plasma samples. Each laboratory used its usual procedures. Mean interlaboratory coefficients of variation (CVs) for the eight plasma samples were 35%, 43%, and 52% for 25-hydroxyvitamin D, 24,25-dihydroxyvitamin D, and 1,25-dihydroxyvitamin D, respectively. Average CVs for the standard solutions were 27%, 23%, and 25%, respectively. Of the eight plasma samples, five had the same concentration for one of the metabolites. One sample was diluted to 0.6 times its original concentration and three samples were fortified with one or more of the metabolites under investigation. Fourteen of 18 laboratories (78%) could distinguish between the five unchanged samples and the modified ones with their 25-hydroxyvitamin D assay. Nine of 12 (75%) could distinguish the modified samples from the other samples with the 24,25-dihydroxyvitamin D assay. Only eight of 15 (53%) could do this their 1,25-dihydroxyvitamin D assay. Values from different laboratories evidently cannot be intercompared without making an actual comparison of the assay procedures. Furthermore, in case of clinical applications of these assays, each laboratory should establish its own reference values and should continually use an internal reference sample to assess the precision of the procedures.