Homeostasis model assessment (HOMA) provides indices of insulin secretion (β) and insulin resistance (R) derived from fasting plasma glucose (FPG) and fasting plasma insulin (FPI) levels. However, these indices could not account for a significant heritability of fasting plasma glucose (FPG) (h2 = 0.75, P < 0.01) in a group of 214 female twins. This result is consistent with a misclassification between effects due to insulin secretion and resistance in the HOMA indices. We report here evidence of such misclassification in the HOMA indices and describe a minor modification to the model which corrects it. Direct measures of insulin resistance (euglycaemic clamp) and secretion (i.v. glucose bolus) were obtained in 43 non-diabetic subjects. Heritability was estimated by statistical modelling of genetic and environmental influences in data from 214 non-diabetic female subjects. Modified HOMA (HOMA′) indices were obtained from β′ = (Ln(FPI)–c)/FPG and R′ = (Ln(FPI)–c)* FPG where c is a constant derived from regression analysis of Ln(FPI) vs FPG. Indices from both models correlated with the direct measures similarly (r = 0.63 (R), 0.49 (R′), 0.45 (β), 0.39 (β′), all P < 0.01). Directly measured insulin resistance and secretion were not significantly correlated (r = 0.13, P = 0.21). However, unmodified HOMA- and R were strongly related (r = 0.78, P < 0.0001 vs 0.13) demonstrating substantial misclassification. The relationship between β′ and R′ (r = 0.13) was not different from that between the two direct measures and significant heritability of β′ (h2 = 0.68, P < 0.01) and R′ (h2 = 0.59, P < 0.05) was evident in the twin data. The proposed modification to HOMA significantly reduces misclassification and reveals separate components of insulin resistance and insulin secretion in the heritability of FPG. Twin Research (2000) 3, 148–151.