Previous studies have indicated that simultaneous infusions of two ketone body tracers ([13C]acetoacetate and [14C]beta-hydroxybutyrate) provide accurate estimates of exogenous ketone body inflow when an open two-pool model is employed. In the present studies, net hepatic ketone body production was determined from surgically placed arterial, portal venous, and hepatic venous catheters in conscious diabetic (n = 6) and 4-day fasted (n = 7) dogs. [13C]acetoacetate and [14C]beta-hydroxybutyrate were infused simultaneously, and ketone body production was calculated from either acetoacetate (AcAc) single-isotope data, beta-hydroxybutyrate (beta-OHB) single-isotope data, the sum of individual fluxes, or the two-pool model. In fasted animals, both the AcAc single-isotope calculation and the sum of individual fluxes overestimated net hepatic production by approximately 50% (P less than 0.05), whereas the beta-OHB single-isotope calculation and the two-pool model gave accurate estimates. In the diabetic animals, the beta-OHB single-isotope calculation underestimated net hepatic production by approximately 30% (P less than 0.05). The sum of individual fluxes overestimated net hepatic production by approximately 46% (P less than 0.05), whereas both the AcAc single-isotope calculation and the two-pool model gave accurate estimates. In conclusion, single-isotope methods give erroneous estimates of net hepatic production of ketone bodies. In contrast, a two-pool model provided an accurate estimate of net hepatic production and thus appears to be suitable for determination of ketone body kinetics in humans.