We measured serum bone alkaline phosphatase (B-ALP) with a new immunoradiometric assay (IRMA) in a large sample of healthy controls comprising 173 women and 180 men, 20-88 yr of age, and in patients with metabolic bone disease. Using serum samples from patients with liver disease and patients with Paget's disease with elevated total alkaline phosphatase (T-ALP) as a source of, respectively, liver and bone isoenzymes, we determined a liver cross-reactivity of the IRMA of 16% that was confirmed by electrophoresis of the circulating alkaline phosphatase isoenzymes. The IRMA was linear for serial sample dilutions, the recovery ranged from 89-110%, and the intra- and interassay variations were below 7% and 9%, respectively. B-ALP increased linearly with age in both sexes, and the mean B-ALP serum levels were not significantly different for women and men (11.3 +/- 4.8 ng/mL for women; 11.0 +/- 4.0 ng/mL for men). The increase in B-ALP after the menopause was significantly higher than that in T-ALP (+77% vs. +24%; P < 0.001). When the values of postmenopausal women were expressed as the SD from the mean of premenopausal women, the mean Z scores were 2.2 +/- 1.8 for B-ALP and 0.9 +/- 1.3 for T-ALP (P < 0.001 between the two). Serum B-ALP was increased from control values in patients with Paget's disease (n = 57; mean, 171.8 +/- 135.6 ng/mL; P < 0.001), in patients with primary hyperparathyroidism (n = 18; mean, 17.2 +/- 5.9 ng/mL; P < 0.001), and in patients with chronic renal failure on hemodialysis (n = 83; mean, 36.6 +/- 35.7 ng/mL; P < 0.001). In patients with Paget's disease, B-ALP was highly correlated with T-ALP (r2 = 0.94; P < 0.001), and the decrease in its serum level was larger than that in T-ALP after treatment with the bisphosphonate pamidronate (-58% vs. -43%; P < 0.03). In patients with various liver diseases, B-ALP was slightly increased, but stayed within the normal range (mean +/- 2 SD) until T-ALP did not exceed 4.5 mu katal/L. We conclude that this new IRMA for B-ALP is reliable, has a low cross-reactivity with the liver isoenzyme, and appears to be more sensitive than T-ALP for the clinical investigation of patients with osteoporosis and other metabolic bone diseases.