Elevated rates of basal hepatic glucose output (bHGO) are significantly correlated with the fasting serum glucose (FSG) level in subjects with non-insulin-dependent diabetes mellitus (NIDDM). This observation suggests that bHGO is a major determinant of the severity of the diabetic state in these subjects. In addition, basal glucagon levels (bGL) are higher in these diabetics than in control subjects, despite the concurrent basal hyperglycemia and hyperinsulinemia, two factors known to suppress glucagon secretion. Although bGL is responsible for sustaining two-thirds of bHGO in normal humans, its role in sustaining elevated rates of bHGO in NIDDM has not been previously defined. To this end, we have studied 13 normal and 10 NIDDM subjects; mean FSG levels were 90 ± 2 and 262 ± 21 mg/dl, respectively (P <.001). The mean fasting serum insulin and glucagon levels were higher in the diabetics than in the controls: 17 ± 2 vs. 9 ± 1 μU/ml (P < .01) and 208 ± 37 vs. 104 ± 15 pg/ ml (P < .01), respectively. On separate days, HGO was assessed isotopically (with 3-[³H]glucose) in the basal state and during infusion of somatostatin (SRIF) (600 μg/h) alone and in conjunction with replacement infusions of glucose and insulin. The results demonstrate that 1) bHGO is higher in diabetics than in controls (145 ± 12 vs. 89 ± 3 mg m² ml⁻¹, P < .01); 2) during infusion of SRIF alone, HGO was suppressed by 25% (P < .05) and 34% (P < .05) of the basal value in controls and diabetics, respectively; 3) when the studies were repeated with glucose levels held constant at or near the FSG level by the glucose-clamp technique, the pattern and degree of HGO suppression was similar to that obtained by infusion of SRIF alone; 4) during isolated glucagon deficiency (SRIF + insulin, 5 mU m⁻² min⁻¹ with serum glucose maintained at basal level), HGO was suppressed by 71 ± 8% of the basal value in controls (P < .001) and by 58 ± 7% in diabetics (P < .001); and 5) when isolated glucagon deficiency with similar hyperglycemia was created in control subjects, HGO was suppressed by 87% of the basal value. We conclude that 1) elevated glucagon levels contribute significantly to the elevated rates of bHGO in NIDDM subjects, 2) basal glucagon levels sustain ∼71 % of basal HGO in control and 58% in NIDDM subjects, and 3) the excess glucagon effect is largely responsible for the apparent hepatic insulin resistance seen in NIDDM subjects.