The effects of exercise on the molecular nature of secreted human growth hormone (GH) or its biological activity are not well understood. Plasma from women (average age 23.6 yr, n = 35), drawn before and after an acute heavy resistance exercise test, was fractionated by size exclusion chromatography into three size classes, namely, > 60 kDa (fraction A), 30-60 kDa (fraction B), and < 30 kDa (fraction C), before GH assay. Concentrations of GH in these fractions, as well as in unfractioned plasma, were measured by the Nichols immunoradiometric assay, National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) polyclonal competitive RIA, Diagnostic Systems Laboratory's immunofunctional assay (measures dimerization-capable species), and the rat tibial bioassay. Significantly increased circulating GH concentrations of two- to fourfold were observed when immunoassays in unfractionated plasma samples were used, but they showed no significant change with use of the rat tibial bioassay. Significant exercise-induced increases in GH were found in fractions B and C but not in fraction A. Because chemical reduction of the samples before GH immunoassay significantly increased GH concentrations in fractions B and C (Nichols and NIDDK kits) after exercise, it is concluded that exercise may specifically increase release of disulfide-linked hormone molecules and/or fragments. Finally, because most of the GH released after exercise was able to dimerize the GH receptor in vitro, it is also concluded that these forms have the two intact binding sites required to initiate signal transduction in target cells.