Little is known about the physiological secretion of the free beta-subunit of LH (LHbeta). The aim of this study was to compare in women the secretion of LHbeta, using sensitive and specific two-site immunoassays, with dimeric LH and the free common alpha-subunit (FAS). The LHbeta assay does not recognize the dimeric LH and cross-reacts only with free hCG beta-subunit (CGbeta). Thus, all of the plasma samples were also tested with a highly specific immunoradiometric assay for free CGbeta. Molar concentrations (i.e. picomoles per L) were used to compare the plasma levels of LH and its free subunits. Plasma LH, LHbeta, FAS, and CGbeta levels were measured in five normally cycling women during the early follicular phase and the ovulatory peak of LH. The pulsatile profiles of LH, LHbeta, FAS, and CGbeta were studied in five postmenopausal women before and 21 days after injection of a depot preparation of the GnRH agonist D-Trp6 (3.75 mg, im) and in five women with functional hypothalamic amenorrhea (FHA), i.e. low plasma LH levels, during pulsatile GnRH administration (20 microg/pulse, 90 min, sc). Afterward, one of the patients with FHA received a single sc injection of 1350 U recombinant human LH, and plasma LH, LHbeta, FAS, and CGbeta levels were measured and compared with the high plasma levels of one postmenopausal woman. In cycling women, basal plasma LHbeta and CGbeta levels were below the detection limit of the assays (1.34 and 0.65 pmol/L, respectively), and plasma FAS levels were 13.60 +/- 0.13 pmol/L. During the LH surge, there was a parallel increase in LH, LHbeta, and FAS. Plasma CGbeta levels remained undetectable. In normal postmenopausal women, basal plasma dimeric LH, LHbeta, and FAS levels were increased in parallel, and their pulsatile profiles were similar, without measurable plasma CGbeta levels. After D-Trp6 administration, plasma LH and LHbeta levels were completely suppressed, whereas plasma FAS levels increased, and plasma CGbeta remained below 0.65 pmol/L. In FHA women, basal plasma levels of LH and FAS were low, without detectable LHbeta and CGbeta levels. During pulsatile GnRH administration, LHbeta became detectable, and pulses were synchronous with those of LH and FAS. The secretion of LH and LHbeta was almost equimolar. Plasma CGbeta levels remained undetectable. In the patient with FHA, administration of recombinant human LH increased only plasma LH levels, whereas plasma LHbeta and FAS levels remained very low. In conclusion, when the production of dimeric LH increases, a concomitant, parallel, and almost equimolar hypersecretion of uncombined and biologically inactive LHbeta occurs. Like the alpha-subunit, LHbeta may be secreted in the dissociated free form. This can lead to pitfalls during clinical investigations if assays of free CGbeta display some cross-reaction with free LHbeta.