Insulin stimulates muscle glucose uptake and inhibits hepatic glucose production. Measures of insulin sensitivity or insulin resistance must take both of these actions into account. Homeostatic model assessment of insulin resistance (HOMA(IR)) and quantitative insulin sensitivity check index (QUICKI) are two simple measures of insulin resistance and insulin sensitivity that are derived from fasting glucose and insulin levels as such it is likely that they reflect primary hepatic insulin action and not muscular effects. To prove this hypothesis, the relationships of HOMA(IR) and QUICKI to peripheral insulin sensitivity (S(I)(*)) and hepatic insulin resistance (HIR) were assessed in 34 adolescents (age 13.5 +/- 2.9 yr, body mass index 23.0 +/- 5.7, mean +/- SD; 16 Caucasian and 14 African-American). S(I)(*) and HIR were determined using the stable-labeled, frequently sampled intravenous glucose tolerance test (250 mg total glucose/kg, 13% [6,6]-D(2)-glucose). The former was calculated using the one-compartment minimal model and stable glucose concentration (S(I)(*)), and the latter was determined over the last hour of the study by multiplying hepatic glucose production (Steele's equation) by mean plasma insulin concentration. As expected, HOMA(IR) and QUICKI were significantly related to HIR (log HOMA(IR) and log HIR, r = 0.73, p < 0.001; QUICKI and log HIR, r = -0.69, p < 0.001) but not S(I)(*). When both S(I)(*) and HIR were included in the equations, only the relationships to log HIR were significant (log HOMA(IR), p < 0.001; QUICKI, p < 0.001). The relationships were similar in African-American and Caucasian subjects. These results demonstrate that in adolescents, HOMA(IR) and QUICKI assess hepatic but not peripheral insulin action.