Low-rate (6 ml/h) intragastric infusion of stable, isotope-labeled amino acids is commonly used to assess the splanchnic handling of amino acids in humans. However, when used in the postabsorptive state, this method yields unreliable plasma isotopic enrichments, with a coefficient of variation >10%. In this metabolic condition, we confirmed in six subjects that an intragastric infusion of L-[²H₃]leucine at 6 ml/h yields an unreliable isotopic steady state in plasma amino acids with a coefficient of variation of 43 ± 12% (mean ± SD). In five additional subjects, we assessed the effects of 1) increasing the rate of delivery of a leucine tracer in an isotonic plasmalike solution at 240 ml/h into the gastric site, and 2) changing the site of infusion from gastric to duodenal with this same high rate of delivery. In contrast to the gastric route, and regardless of the rate of delivery, only the intraduodenal route allowed 1) isotopic plasma steady state (i.e., coefficients of variation were <10%: 5 ± 3%), and 2) reproducible leucine extraction coefficients (22 ± 5%). We conclude that an infusion site that bypasses the gastric emptying process, i.e., the duodenal route, along with delivery of a plasmalike solution, is necessary to reach isotopic steady state in plasma when labeled leucine is infused into the gastrointestinal tract in the postabsorptive state.