Three monoclonal antibodies recognizing the 5-13, 25-37, and 43-49 sequence of the human osteocalcin were used in competitive and two-site radioimmunoassays (RIA) to characterize specifically various immunoreactive forms of circulating human osteocalcin. The intact molecule accounts for 36% of total in normals (2.6 nM), 46% in patients with osteoporosis (3.1 nM), and 26% in chronic renal failure (6.9 nM). Four fragment were detected in addition to the intact molecule in the serum of healthy adults and patients with metabolic bone disease. N-terminal, mid, and mid C-terminal fragments were present in minute amounts (each accounting for 5-14% of the total circulating osteocalcin immunoreactivity). In contrast, the N-terminal midfragment, probably resulting from the cleavage around amino acids 43-44, represents about 30% (2 nM) of the total osteocalcin immunoreactive level in normals and patients with osteoporosis and up to 50% (13 nM) in patients with chronic renal failure. This large N-terminal midfragment, representing 75-80% of the intact osteocalcin level, is not lower when the plasma assay is performed immediately after sampling (within 20 minutes at 4 degrees C with proteinase inhibitors), indicating that it circulates in vivo. In addition, this fragment was detected in the supernatant of osteoblastic cells, representing about 28% of the intact peptide. Levels of N-terminal midfragment were not changed after treatment of patients with metabolic bone disease (Paget's disease, reflex sympathetic dystrophy, fibrous dysplasia, and osteoporosis) by bisphosphonate, suggesting that it is not released during bone resorption. The osteocalcin level measured with the two-site immunoradiometric assay specific for the intact molecule or with a conventional bovine RIA was rapidly decreased after incubation of serum at room temperature (-20 and -15%, respectively, after 3 h), whereas the total level of intact osteocalcin plus N-terminal midfragment was not changed. Intact osteocalcin loss can be partially avoided by proteinase inhibitors and by incubating serum at 4 degrees C. In conclusion, we characterized multiple immunoreactive forms of osteocalcin that circulate in addition to the intact molecule, none of them being specifically altered in osteoporosis. The N-terminal midfragment circulates in a large amount, probably resulting from cleavage of the intact molecule in the circulation and/or at peripheral sites. These fragments can also be generated in vitro by proteolytic degradation of the intact molecule. To obtain reliable intact osteocalcin values but also reliable levels measured with conventional competitive RIA, careful control of the sampling conditions is warranted.