A mathematical model of normal glucose/insulin homoeostasis has been based on the known, experimentally determined responses of the liver and periphery to different glucose/insulin concentrations. Different defects of glucose resistance and insulin resistance have been applied to the model to investigate the degree to which these abnormalities could successfully predict the range of fasting glucose and insulin values found in normal, obese, and diabetic subjects. Modeling glucose resistance or insulin resistance at the liver or the periphery alone did not increase the plasma glucose to levels observed in diabetes, even when associated with marked deficiency of beta-cell function. A defect of either glucose resistance or insulin resistance affecting both periphery and liver allowed a wider range of basal glucose and insulin concentration values, but resulted in unphysiologically low hepatic glucose output: with modeling insulin resistance, hyperglycemia suppressed glucose output, whereas with glucose resistance, raised insulin levels suppressed hepatic glucose output. A wide range of glucose and insulin values, with appropriate basal hepatic glucose output, could only be modeled by insulin resistance at both the liver and periphery with additional glucose resistance at the liver. The modeling results are in accord with investigative studies that suggest secondary hepatic and peripheral glucose resistance in response to hyperglycemia. Modeling provides a systematic means of examining the likely effect of different putative defects in a complex physiological system.