Carbohydrate ingestion causes the release of two in-sulinotropic intestinal hormones, glucose-dependentinsulinotropic polypeptide (GIP) [1] and glucagon-like peptide-1 (GLP-1) [2]. GIP is mainly releasedfrom the duodenum, and GLP-1 from the ileum, andthis probably explains why the GIP response to amoderate oral glucose load is large whereas theGLP-1 response is very small [3]: the glucose is ab-sorbed before it reaches the GLP-1 secreting cells.The augmented secretion of insulin resulting fromthe insulinotropic actions of the two hormones, theso-called incretin effect [4, 5], is thought to be impor-tant for maintaining a normal glucose tolerance. ForGLP-1, this has recently been confirmed with the aidof a GLP-1 receptor antagonist, exendin 9–39, the ad-minstration of which greatly reduced the insulin re-sponse to intestinal glucose in rats [6, 7]. Further-more, targeted disruption of the GLP-1 receptor inmice resulted in glucose intolerance and basal hyper-glycaemia (at least in male animals) [8]. GIP has asimilar, well documented [9] incretin effect and, in arecent study administration of a GIP receptor antag-onist reduced postprandial insulin release in con-scious rats by 72% [10]. It seems probable, therefore,that elimination of its insulinotropic actions wouldalso result in glucose intolerance.A feature that characterises patients with NIDDMis an absent or grossly impaired incretin effect, evi-dent after administration of small to moderateamounts of glucose [11–13]. A number of studieshave addressed secretion of GIP and GLP-1 in dia-betes, but most agree that, if anything, secretion,particularly of GIP, is increased rather than impairedin NIDDM [12, 14]. Secretory defects, therefore, donot explain this decreased incretin effect. Further-more, there is no evidence that the structure of thehormones is defective in NIDDM. However, whenone compares the effects of the two hormones on in-sulin secretion in NIDDM patients, an interesting dif-ference emerges [15, 16]. GLP-1 is strongly insulino-tropic in these patients. Indeed, in patients with mod-erate disease, insulin secretion (as estimated by C-peptide responses) was near normal when comparedto that of healthy control subjects under the condi-tions of an identical hyperglycaemic clamp [15]. Thepeptide is so effective that its i.v. administration maycompletely normalise blood glucose levels [17]. Thishas been observed in all patients investigated in thislaboratory so far, regardless of the severity or dura-tion of disease. Part of this remarkable effect is un-doubtedly due to the concomitant inhibition of gluca-gon secretion [17, 18] and, as for meal-induced glu-cose excursions, its inhibitory effect on gastric empty-ing [19], but in all patients, the hormone causes a sig-nificant release of insulin. In fact, GLP-1 is the mostefficacious and potentinsulinotropichormoneknown[2, 4, 20]. Possibly, with GLP-1, one may actuallygauge the secretory capacity of the residual beta-cellmass in patients with diabetes mellitus. Why, then, isthe incretin effect lost in NIDDM patients? It is typi-cal of the studies reporting a lack of incretin effect,that small or moderate oral glucose loads were ad-ministered. As already mentioned, small glucoseloads represent a very weak stimulus for GLP-1 se-cretion. Presumably, the incretin effect of GLP-1 ispreserved in NIDDM, but this is overlooked withconventional incretin analysis, simply because it doesnot bring about appropriate stimulation of GLP-1 se-cretion [3]. In contrast, it turns out that GIP has littleor no effect on insulin secretion in these patients!This has been found in studies employing both