The basic tenet of this investigation was that obesity is not a prerequisite in the development of polycystic ovary syndrome (PCOS), as indicated by the fact that 50% of PCOS women are not obese. Further, obesity itself is a disease entity with the common manifestation of insulin resistance/hyperinsulinemia with PCOS. Given recent evidence that insulin and GH may have gonadotropin-augmenting effects, we have determined the common and distinguishing features of neuroendocrine-metabolic dysfunctions of lean [body mass index (BMI), < 23 kg/m2] and obese (BMI, > 30 kg/m2) women with the classical form of PCOS. Insulin sensitivity, as determined by rapid i.v. glucose tolerance testing; 24-h dynamics of insulin/glucose levels, somatotropic [GH/GH-binding protein/insulin-like growth factor I (IGF-I)/IGF-binding proteins (IGFBP)], and LH axes; and their downstream effects on ovarian steroids were simultaneously assessed in eight lean PCOS and eight obese PCOS patients and an equal number of BMI-matched normal cycling controls. Our results show that insulin sensitivity was reduced 50% (P < 0.01) in lean PCOS from that in lean controls. There was a further decrease in obese controls (P < 0.01) and a 2-fold greater reduction (P < 0.001) in obese PCOS than in obese controls, suggesting that insulin resistance (IR) is a common lesion in PCOS, and that obesity contributes an additional component to IR in obese PCOS. Consistent with the degree of IR, the manifestation of compensatory hyperinsulinemia in lean PCOS was incipient, being evident only in response to meals (P < 0.05), and became overt during the 24-h fasting/feeding phases of the day in obese control (P < 0.001) with a 2- to 3-fold greater elevation (P < 0.001) in obese PCOS. An enhanced early insulin response to glucose occurs equally in obese control (P < 0.01) and obese PCOS (P < 0.05), but not in their lean counterparts. Considering the more profound IR and the associated hyperglycemia in obese PCOS, the magnitude of the early insulin release is inadequate, suggesting that beta-cell dysfunction exists in obese PCOS. Remarkable differences in the somatotropic axis were also observed; although 24-h GH pulse frequency and levels of IGF-I and IGFBP-3 were unaltered by either PCOS or obesity, the 24-h mean GH pulse amplitude was increased by 30% (P < 0.01) in lean PCOS in the presence of normal levels of high affinity GHBP and normal GH response to GHRH. In distinct contrast, the somatotropic axis in both obese control and obese PCOS was profoundly modified, with attenuation of GH pulse amplitude (P < 0.001) and GH response to GHRH (P < 0.001), resulting in a state of hyposomatotropinism with a more than 50% reduction (P < 0.001) of 24-h mean GH levels. In addition, GHBP levels were elevated 2-fold and were correlated inversely with GH (r = -0.81) and positively with insulin (r = 0.75) concentrations. IGFBP-I levels were suppressed in both obese groups, with a 4-fold greater reduction in obese PCOS than that in obese controls. Thus, the downstream effects of hyperinsulinemia on the somatotropic axis may include up-regulation of hepatic production of GHBP, suppression of IGFBP-1 (r = 0.82) and sex hormone-binding globulin (r = -0.69) levels, and a more than 3-fold increase in ratios of IGF-I/IGFBP-1 and estradiol-testosterone/sex hormone-binding globulin, thereby increasing their bioavailabilities. In contrast, LH pulsatility was unaffected by obesity alone. An accelerated LH pulse frequency was evident in both lean and obese PCOS (P < 0.001), whereas the mean 24-h LH pulse amplitude was increased in lean (P < 0.001), but not obese, PCOS patients. These events resulted in a 3-fold increase in 24-h mean LH levels in lean PCOS and a 2-fold increase in obese PCOS. Thus, increased LH pulse frequency and augmented LH response to GnRH are characteristic of PCOS, independent of obesity, and the presence of obesity in PCOS is associated with an attenuated LH pulse amplitude, not accounted f