We compared the clinical performance of a carboxyl-terminal radioimmunoassay for human parathyroid hormone (iPTH), using either a dynamic reference interval (95% confidence limits of serum iPTH concentrations observed in 11 normal individuals during intravenous infusions of Na2EDTA and CaCl2) or a gaussian (2 SD) reference interval derived from 233 normocalcemic individuals. The 2 SD ranges were 3.5 to 9.8 pmol/L for serum iPTH and 2.19 to 2.53 mmol/L for total calcium. The iPTH dynamic interval was lower for calcium concentrations greater than 2.50 mmol/L; it was higher, wider, and continued to increase for calcium values less than or equal to 2.25 mmol/L. Use of the dynamic reference interval increased the clinical sensitivity of our assay from 81% and 61% to 100%, respectively, in primary hyperparathyroidism (n = 47) and hypoparathyroidism (n = 18). Test specificity was maintained at 100% in hypocalcemic disorders but fell to 93% (62/67) in hypercalcemic disorders. Overall, use of the dynamic reference interval improved the assay performance.