In the male rodent and primate, fasting or severe caloric restriction significantly decreases serum testosterone concentrations, putatively via inducing secondary hypogonadotrophism. To clarify this presumptive pathophysiology, we have used: 1) a high sensitivity immunoradiometric assay, which correlates well with an in vitro Leydig cell bioassay of LH; 2) blood sampling every 5 min for 24 h basally and every 10 min for 3 h after GnRH injection before and after a 5-day (water only) fast in eight healthy young men; and 3) deconvolution analysis to evaluate in vivo LH secretory burst frequency, amplitude, duration, and mass, and LH half-life simultaneously. We documented a 50% fall in serum total and free testosterone concentrations, and a 30% decrease in 24-h mean serum LH concentrations (viz., fed 3.0 +/- 0.47 vs. fasted 2.1 +/- 0.39 U/L, P = 0.043). Deconvolution analysis revealed preservation of LH secretory pulse frequency (fed 12.9 +/- 0.48 vs. fasted 12.6 +/- 0.78 secretory bursts/day, P = NS) during fasting-induced hypogonadotropism. The duration of computer-resolved LH secretory bursts, the interburst interval, and the calculated endogenous half-life of LH also did not change, whereas LH secretory burst mass declined significantly; viz. from 28 +/- 5 in the fed to 14 +/- 3.2 U/L of distribution volume/day in the fasted state (P = 0.034). In contrast, LH release after a 10 micrograms pulse of GnRH iv was enhanced during fasting in seven of the eight men. Fasting also decreased mean (24 h) serum TSH and PRL, increased cortisol, dehydroepiandrosterone sulfate and GH, and did not affect FSH concentrations or the radioiodinated albumen distribution space. In summary, in young men 5 days of nutrient deprivation selectively attenuates the mass of LH secreted per burst without altering LH secretory event frequency or LH half-life. We infer that decreased LH release per burst is due to decreased hypothalamic GnRH impulse strength, since LH release induced by a submaximally effective pulse of exogenous GnRH is amplified rather than attenuated.