In ruminants, the main precursor for GNG is ruminal propionate which fundamentally differs from gluconeogenic precursors of monogastrics. The D₂O method has been used to measure fractional gluconeogenesis (GNG) in humans. We aimed to determine whether the route of D₂O administration (rumen (IR) or jugular vein (IV)) affects D enrichment of rumen fluid, ruminal propionate and estimates of fractional GNG. In addition, we compared these GNG estimates with estimates derived from ¹³C₆-glucose infusion. Four sheep (23.5 ± 1 kg BW), equipped with a rumen fistula and a jugular vein catheter, were fed a pelleted ration (35 g/kg BW and d; 9 MJ ME/d) at 2-h intervals. Water was offered ad lib. Sheep were given 2 boli of 7 g D₂O (99.2 atom% (AP) D/kg BW) at 0800 and 1200 h either IR (method 1) or IV (method 2), or received continuous IV infusion of ¹³C₆-glucose (0.15 g/h) for 10 h (method 3) in a crossover design with 1 wk separating tracer administrations. Ruminal D enrichments were measured by isotope ratio mass spectrometry whereas ruminal propionate and plasma glucose D enrichment was measured by GC-MS. Fractional GNG was calculated from the ratio of D enrichment at C-5 of glucose (labeling via GNG only) to that at C-2 (labeling during GNG and glycogenolysis). The ¹³C-mass isotopomer enrichment of plasma glucose was determined by GC-MS and GNG was calculated. Statistical comparison of GNG estimates was made with repeated measures ANOVA using PROC MIXED of SAS. Rumen fluid D enrichment attained a plateau 6 h after the first bolus (IR: 1.51; IV: 1.43 APE; P > 0.1). The D enrichment of ruminal propionate as a precursor for GNG showed faster labeling with the IR route (3 h: P < 0.10; 6 h: P < 0.05). However, the plateau enrichments of rumen fluid D and propionate D did not differ (P > 0.10). GNG estimates derived from IR and IV routes of D₂O administration did not differ (P = 0.83) which resulted in an overall GNG estimate of 58.8%. In contrast, GNG estimate derived from ¹³C₆ glucose dilution was at 72.8%, which did not differ from the other 2 methods (P > 0.13). Thus all 3 methods yield similar estimates of GNG in ruminants.