The inability to detect insulin-like growth factor binding protein-3 (IGFBP-3) in some circumstances by Western ligand blot analysis has emphasized the need to characterize IGFBPs by both ligand binding and immunological techniques. In this study, we have: 1) characterized and quantified IGFBP-3 in nonpregnancy, pregnancy, and fetal cord serum, follicular, peritoneal, and amniotic fluid, seminal plasma, cerebrospinal fluid (CSF), and urine; 2) established a new IGFBP-3 RIA that detects both intact and fragments of IGFBP-3; 3) identified both intact and fragments of IGFBP-3 by Western immunoblot techniques; and 4) addressed the discordance between Western ligand blot analysis and RIA by assessing fluids for IGFBP proteolytic activity. All fluids examined, except pregnancy serum, CSF, and amniotic fluid, displayed a 44-34-kilodalton (kDa) IGFBP-3 doublet by Western ligand blot analysis. Western immunoblot analysis using specific IGFBP-3 antiserum showed a 44-34-kDa IGFBP-3 doublet and a 28-kDa fragment in nonpregnancy serum, fetal cord serum, follicular fluid, and peritoneal fluid. The immunoreactive 42-38-kDa doublet was faint in urine and seminal plasma. IGFBPs in CSF did not cross-react with IGFBP-3 antiserum. Pregnancy serum and amniotic fluid contained only the 28-kDa fragment when compared against equal volumes of nonpregnancy serum. With the development of an IGFBP-3 RIA, IGFBP-3 could be accurately measured; urine, CSF, and seminal plasma contained the lowest levels of IGFBP-3 at 27 +/- 3 ng/ml (mean +/- SEM), 110 +/- 26 ng/ml, and 209 +/- 56 ng/ml, respectively. In increasing concentration: fetal cord serum contained 753 +/- 101 ng/ml; peritoneal fluid, 1124 +/- 130 ng/ml; follicular fluid, 2356 +/- 211 ng/ml; nonpregnancy serum, 3556 +/- 508 ng/ml; pregnancy serum, 3718 +/- 842 ng/ml; and amniotic fluid, 5150 +/- 688 ng/ml. The measurable concentrations of IGFBP-3 in CSF and the high concentrations measured in pregnancy serum and amniotic fluid conflicted with Western blot analysis. Thus, fluids were assessed for IGFBP proteolytic activity by incubation with a source of IGFBP-3, either nonpregnancy serum or purified IGFBP-3. All fluids displayed some proteolytic activity with either assay. Fluids with little protease activity (nonpregnancy serum, follicular fluid, and urine) showed a close relationship between immunoassayable IGFBP-3 by RIA and IGFBP-3 band intensity by Western ligand blot. Fluids with high proteolytic activity (pregnancy serum, CSF, seminal plasma, peritoneal fluid, and amniotic fluid) gave discrepant IGFBP-3 values between RIA and Western ligand blot.(ABSTRACT TRUNCATED AT 400 WORDS)