Women with polycystic ovary syndrome (PCOS) are profoundly insulin resistant, and the resultant hyperinsulinemia exacerbates the reproductive abnormalities of the syndrome. Agents that ameliorate insulin resistance and reduce circulating insulin levels could provide a new therapeutic modality for PCOS. Identifying the subset of PCOS women who are most insulin resistant may therefore be useful for selecting women who will respond to this therapy. We examined the correlation of basal and oral glucose-stimulated glucose and insulin levels and fasting and stimulated glucose/insulin (G:I) ratios with parameters of insulin sensitivity obtained by frequently sampled i.v. glucose tolerance test (FSIGT) to assess whether there is a simple screening test for insulin resistance in PCOS. Forty PCOS women (aged 18-40 yr; body mass index, >26 kg/m2) and 15 control women matched for age, weight, and ethnicity underwent both a 75-g oral glucose tolerance test (OGTT) and a FSIGT. The insulin sensitivity index (S(I)) was calculated by application of the minimal model of glucose kinetics to the dynamics of plasma glucose and insulin levels during the FSIGT. The best correlation in PCOS between S(I) and a fasting level was found with fasting G:I ratios (r = 0.73; P < 0.0001). A less substantial, but significant, correlation was found with fasting insulin levels (r = 0.50; P < 0.001), and no significant correlation was found with fasting glucose levels (r = 0.24; P = NS). The fasting G:I was more strongly correlated with S(I) than with integrated glucose and insulin responses during the OGTT. The only stronger correlation was with the OGTT 2 h G:I ratio (r = 0.74; P < 0.001). Stepwise regression analysis with S(I) as the dependent variable and fasting glucose and insulin levels, area under the curve for glucose and insulin, and a fasting G:I ratio showed that only the fasting G:I ratio was significantly predictive of S(I) in the model (F to remove value = 38.1; P < 0.001). When viewed as a screening test for insulin resistance in PCOS, setting a value of the fasting G:I ratio of less than 4.5 as abnormal (using an S(I) value below the 10th percentile of our control population as evidence for insulin resistance), the sensitivity of a fasting G:I ratio was 95%, the specificity was 84%, the positive predictive value was 87%, and the negative predictive value was 94%. Receiver operator curve analysis showed that this fasting G:I ratio was the single best screening measure for detecting insulin resistance. We conclude that a fasting G:I ratio may be useful as a screening test for insulin resistance in obese non-Hispanic white PCOS women. This may be a clinically useful parameter for selecting PCOS women most likely to respond to therapeutic interventions that improve insulin sensitivity.