Relationships were determined between the labeling of leucine and phenylalanine at the intracellular site of protein synthesis in the pancreas and labeling of plasma leucine, its keto acid, alpha-ketoisocaproate (KIC), and phenylalanine. Six healthy subjects were studied for 480 minutes during a primed constant infusion of [1-14C]leucine, and six healthy subjects were studied for 240 minutes with [1-14C]leucine, [4,5-3H]phenylalanine, and [1-13C]KIC. An oro-duodenal tube was placed and pancreatic exocrine fluid was sampled by duodenal aspiration during cholecystokinin stimulation. During the 480-minute study, in the final 120 minutes the specific activity (SA) of enzyme leucine (3.14 +/- 0.27 dpm/nmol) was lower than that of plasma leucine (4.18 +/- 0.30 dpm/nmol, P < .001), but was not different from that of plasma KIC (3.02 +/- 0.18 dpm/nmol). During the 240-minute study, protein synthesis rates of secreted pancreatic enzymes when calculated with [3H]phenylalanine were lower (P = .006) by 28% +/- 2% than rates based on [14C]KIC SA, and lower (P = .004) by 16% +/- 3% than those calculated using [14C]leucine SA. Incorporation of [13C]leucine into pancreatic enzymes was not different from that of [14C]leucine when [13C]leucine and [14C]KIC, respectively, were used to denote precursor labeling. The results indicate that plasma KIC SA reflects the precursor pool for pancreatic protein synthesis during leucine tracer infusion, and plasma leucine enrichment also reflects the precursor pool when [1-13C]KIC is infused in man. The precursor pool is erroneously overestimated when using plasma SA of [4,5-3H]phenylalanine or [1-14C]leucine.