To improve upon the [U-(13)C]glucose method to estimate "gluconeogenesis" as described by J. Katz and J. A. Tayek (Am. J. Physiol. Endocrinol. Metab. 272: E476-E484, 1997, and 275: E537-E542, 1998), we describe the reciprocal pool model by using only the isotopomer data of plasma glucose during infusion of [U-(13)C]glucose. The glucose pool serves as both precursor and product for the calculation of the fraction of molecules generated by gluconeogenesis and to correct for exchange and loss of labeled carbon at the level of the tricarboxylic acid cycle. We have applied this model to both our own data and those of other investigators using [U-(13)C]glucose and have demonstrated excellent agreement between the Katz and Tayek model and our reciprocal pool model. When we compare the results of the reciprocal pool model with those of Hellerstein ([2-(13)C]glycerol) and Landau ((2)H(2)O-glucose-C-5), the results are similar in short- and long-term fasted adult humans. Finally, when we apply the reciprocal pool model to our data from premature infants, it is clear that we account for the inflow of unlabeled glycerol and presumably amino acids. This is not surprising, because the vast majority of gluconeogenesis is the result of recycling of glucose and pyruvate carbon.