Insulin generates/releases chemical substances that alter the activities of purified enzymes [1,2] and the metabolism of intact cells [3,4] in an insulin mimetic fashion. These substances have therefore been termed mediators or second messengers of insulin [5]. Two different insulin mediators have been separated from rat liver [6]. These mediators have been identified as inositol phosphoglycans (IPGs), containing inositol, non-acetylated amino sugars, neutral sugars, ethanolamine, phosphate and, possibly, amino acids [5]. One contains d -chiro-inositol and activates pyruvate dehydrogenase (PDH) phosphatase, whereas another contains myo-inositol and inhibits cyclic AMP-dependent protein kinase (PKA). Injection of each of these IPGs into low-dose streptozotocin-diabetic rats lowers blood glucose [7].Body fluids of patients with non-insulin-dependent diabetes mellitus (NIDDM) have abnormally low levels of Summary Insulin mediators (inositol phosphoglycans) have been shown to mimic insulin action in vitro and in intact mammals, but it is not known which mediator is involved in insulin action under physiological conditions, nor is it known whether insulin resistance alters the mediator profile under such conditions. We therefore investigated the effects of glucose ingestion on changes in the bioactivity of serum inositol phosphoglycan-like substances (IPG) in healthy men and insulin resistant (obese, non-insulin-dependent diabetic) men. Two classes of mediators were partially purified from serum before and after glucose ingestion. The first was eluted from an anion exchange resin with HCl pH 2.0, and bioactivity was determined by activation of pyruvate dehydrogenase in vitro. The second was eluted with HCl pH 1.3, and bioactivity was determined by inhibition of cyclic AMP-dependent protein kinase. In healthy men, the bioactivity of the pH 1.3 IPG was not altered by glucose ingestion, whereas bioactivity of the pH 2.0 IPG increased to approximately 120 % of the pre-glucose ingestion value at 60-240 min post-glucose ingestion (p < 0.05 vs pre-glucose). There was no change in either IPG after glucose ingestion in the insulin-resistant group. These data suggest that the pH 2.0 IPG plays an important role in mediating insulin's effect on peripheral glucose utilization in man under physiological conditions. The data further show, for the first time, a defective change in the bioactivity of an insulin mediator isolated from insulin-resistant humans after hyperinsulinaemia, suggesting that inadequate generation/release of IPGs is associated with insulin resistance. [Diabetologia (1997) 40: 557-563] Keywords Inositol phosphoglycans, non-insulin dependent diabetes mellitus, pyruvate dehydrogenase, cyclic AMP dependent protein kinase.