The insulinotropic intestinal hormone GLP-1 is thought to exert one of its effects by direct action on the pancreatic beta-cell receptors. GLP-1 is rapidly degraded in plasma, such that only a small amount of the active form reaches the pancreas, making it questionable whether this amount is sufficient to produce a direct incretin effect. The aim of our study was to assess, in a dog model, the putative incretin action of GLP-1 acting directly on the beta-cell in the context of postprandial rises in GLP-1 and glucose. Conscious dogs were fed a high-fat, high-carbohydrate meal, and insulin response was measured. We also infused systemic glucose plus GLP-1, or glucose alone, to simulate the meal test values of these variables and measured insulin response. The results were as follows: during the meal, we measured a robust insulin response (52 +/- 9 to 136 +/- 14 pmol/l, P < 0.05 vs. basal) with increases in portal glucose and GLP-1 but only limited increases in systemic glucose (5.3 +/- 0.1 to 5.7 +/- 0.1 mmol/l, P = 0.1 vs. basal) and GLP-1 (6 +/- 0 to 9 +/- 1 pmol/l, P = 0.5 vs. basal). Exogenous infusion of systemic glucose and GLP-1 produced a moderate increase in insulin (43 +/- 5 to 84 +/- 15 pmol/l, 43% of the meal insulin). However, infusion of glucose alone, without GLP-1, produced a similar insulin response (37 +/- 6 to 82 +/- 14 pmol, 53% of the meal insulin, P = 0.7 vs. glucose and GLP-1 infusion). In conclusion, in dogs with postprandial rises in systemic glucose and GLP-1, the hormone might not have a direct insulinotropic effect and could regulate glycemia via indirect, portohepatic-initiated neural mechanisms.