Galactose is incorporated by a different metabolic pathway than glucose. Its contribution to glycogen synthesis has not been studied in humans. We administered galactose (0.5 g/kg iv) to overnight-fasted normal human volunteers and examined its effects on hepatic glycogen synthesis and hepatic glucose output (HGO). Hepatic glycogenesis was assessed noninvasively, determining glycogen concentration by 13C magnetic resonance spectroscopy (MRS) and liver volume by magnetic resonance imaging. HGO was determined by [6,6-2H2]glucose and gluconeogenesis calculated by adding the amount of hepatic glycogenesis to the HGO. After galactose administration, liver glycogen concentration (baseline 254 +/- 11 mmol/l) decreased in the first 45 min to 207 +/- 15 mmol/l (P < 0.05) and increased thereafter to 313 +/- 7 mmol/l (P < 0.01). Net hepatic glycogenesis was 101 +/- 12 mmol over 150 min. HGO (baseline 14.3 +/- 1.9 mumol.kg-1.min-1) increased threefold in the first 15 min and then returned to baseline. The average rate of gluconeogenesis was 12.3 mumol.kg-1.min-1. Intravenous galactose leads to an increase in hepatic glycogen and hepatic glucose output in normal humans. Competitive inhibition of UDP-glucose pyrophosphorylase by UDP-galactose could explain the apparent glycogenolysis observed early after galactose administration. 13C MRS in combination with a stable isotope tracer is a noninvasive and safe method to study hepatic carbohydrate metabolism in humans.