Fluxes through intrahepatic glucose-producing metabolic pathways were measured in normal humans during overnight or prolonged (60 h) fasting. The glucuronate probe was used to measure the turnover and sources of hepatic UDP-glucose; mass isotopomer distribution analysis from [2-13C1]glycerol for gluconeogenesis and UDP-gluconeogenesis; [U-13C6]glucose for glucose production (GP) and the direct UDP-glucose pathway; and [1-2H1]galactose for UDP-glucose flux and retention in hepatic glycogen. After overnight fasting, GP (fluxes in milligram per kilogram per minute) was 2.19+/-0.09, of which 0.79 (36%) was from gluconeogenesis, 1.40 was from glycogenolysis, 0.30 was retained in glycogen via UDP-gluconeogenesis, and 0.17 entered hepatic UDP-glucose by the direct pathway. Thus, total flux through the gluconeogenic pathway (1.09) represented 54% of extrahepatic glucose disposal (2.02) and the net hepatic glycogen depletion rate was 0.93 (46%). Prolonging [2-13C1]glycerol infusion slowly increased measured fractional gluconeogenesis. In response to prolonged fasting, GP was lower (1. 43+/-0.06) and fractional and absolute gluconeogenesis were higher (78+/-2% and 1.11+/-0.07, respectively). The small but nonzero glycogen input to plasma glucose (0.32+/-0.03) was completely balanced by retained UDP-gluconeogenesis (0.31+/-0.02). Total gluconeogenic pathway flux therefore accounted for 99+/-2% of GP, but with a glycogen cycle interposed. Prolonging isotope infusion to 10 h increased measured fractional gluconeogenesis and UDP-gluconeogenesis to 84-96%, implying replacement of glycogen by gluconeogenic-labeled glucose. Moreover, after glucagon administration, GP (1.65), recovery of [1-2H1]galactose label in plasma glucose (25%) and fractional gluconeogenesis (91%) increased, such that 78% (0.45/0.59) of glycogen released was labeled (i.e., of recent gluconeogenic origin). In conclusion, hepatic gluconeogenic flux into glycogen and glycogen turnover persist during fasting in humans, reconciling inconsistencies in the literature and interposing another locus of control in the normal pathway of GP.