The possible diagnostic or prognostic significance of changes in circulating level of endothelins in a variety of pathological conditions is currently of interest. Unfortunately, no consensus regarding optimization of sensitivity and extraction procedures for the reliable radioimmunoassay of endothelin-1 (ET-1), big endothelin-1 (BigET-1), and endothelin-3 (ET-3) currently exists. The object of the present study was to evaluate aspects of currently used extraction and assay procedures that limit accurate determination of ET in human plasma and define criteria to reduce variability. Critical parameters include the selectivity of commercial antibodies and the ability to remove interfering material after Sep-Pak absorption by selective washing with 24% ethanol in 4% acetic acid or methylene chloride in 0.1% trifluoroacetic acid. Assay sensitivity and specificity in the physiological range is improved by optimizing total binding parameters for the antibodies to give approximately 15-20% binding of radiolabeled peptide. With these modifications normal plasma values for ET-1, BigET-1, and ET-3 averaged 1.7 ุ 0.06, 2.5 ุ 0.3, and 5.8 ุ 0.2 pg/ml, respectively. These data suggest that such modifications may help to resolve many of the earlier difficulties concerning the role of ET under normal and pathological conditions.