Sources of hepatic glycogen synthesis during an oral glucose tolerance test were evaluated in six healthy subjects by enrichment of a 75-g glucose load with 6.67% [U-(13)C]glucose and 3.33% [U-(2)H(7)]glucose and analysis of plasma glucose and hepatic uridine diphosphate-glucose enrichments (sampled as urinary menthol glucuronide) by (2)H and (13)C nuclear magnetic resonance. The direct pathway contribution, as estimated from the dilution of [U-(13)C]glucose between plasma glucose and glucuronide, was unexpectedly low (36 +/- 5%). With [U-(2)H(7)]glucose, direct pathway estimates based on the dilution of position 3 (2)H-enrichment between plasma glucose and glucuronide were significantly higher (51 +/- 6%, P = 0.05). These differences reflect the exchange of the carbon 4, 5, and 6 moiety of fructose-6-phosphate and glyceraldehyde-3-phosphate catalyzed by transaldolase. As further evidence of this exchange, (2)H-enrichments in glucuronide positions 4 and 5 were inferior to those of position 3. From the difference in glucuronide positions 5 and 3 enrichments, the fraction of direct pathway carbons that experienced transaldolase exchange was estimated at 21 +/- 4%. In conclusion, the direct pathway contributes only half of hepatic glycogen synthesis during an oral glucose tolerance test. Glucose tracers labeled in positions 4, 5, or 6 will give significant underestimates of direct pathway activity because of transaldolase exchange.