Direct measurement of de novo lipogenesis has not previously been possible in humans. We measured de novo hepatic lipogenesis in normal men by means of stable isotopes and by combining the acetylated-xenobiotic probe technique with mass isotopomer analysis of secreted very low density lipoprotein-fatty acids (VLDL-FA). Sulfamethoxazole (SMX) was administered with [13C]acetate during an overnight fast followed by refeeding with intravenous glucose (7-10 mg/kg of weight per min), oral Ensure (7-10 mg of carbohydrate/kg of weight per min), or a high-carbohydrate mixed-meal breakfast (3.5 g of carbohydrate/kg of weight). Respiratory quotients remained less than 1.0. High-performance liquid chromatography/mass spectrometry-determined enrichments in SMX-acetate attained stable plateau values, and hepatic acetyl-coenzyme A (CoA) dilution rate did not increase with refeeding (approximately 0.024 mmol/kg per min). The fraction of VLDL-palmitate derived from de novo lipogenesis was only 0.91 +/- 0.27% (fasted) and 1.64-1.97% (fed). For stearate, this was 0.37 +/- 0.08% and 0.47-0.64%. Precursor enrichments predicted from isotopomer ratios were close to measured SMX-acetate enrichments, indicating that SMX-acetate samples the true lipogenic acetyl-CoA pool. Stearate synthesis was less than palmitate and the two did not move in parallel. Estimated total VLDL-FA synthesis is less than 500 mg/day. Thus, de novo hepatic lipogenesis is a quantitatively minor pathway, consistent with gas exchange estimates; fatty acid futile cycling (oxidation/resynthesis) is not thermogenically significant; and synthesis rates of different nonessential fatty acids by human liver are not identical in nonoverfed normal men. The contribution and regulation of de novo lipogenesis in other settings can be studied using this technique.