Despite the evidence in experimental animal models that insulin, or GIK (glucose-insulin-potassium), improves left ventricular function and perfusion during both acute and chronic ischaemia, clinical studies have generated conflicting results. We tested the hypothesis that pretreatment with GIK attenuates the vascular and functional effects of stress-induced myocardial ischaemia in humans. Twenty-two patients with evidence of inducible myocardial ischaemia were enrolled; 11 patients with normal ventricular function underwent two dipyridamole echocardiography tests, and 11 with regional contractility defects from previous myocardial infarction were submitted to two ECG exercise tests combined with 201Tl myocardial perfusion scintigraphy; the tests were preceded by 60 min of either normal saline or an isoglycaemic GIK infusion. On a stress echocardiogram, a 30% reduction in the severity of ischaemia was observed. On ECG ergometry, GIK infusion slightly increased the time to ischaemia (+0.6 min, P=0.07); however, the higher workload (+8%, P=0.07) was achieved at a similar rate-pressure plateau. On scintigraphy, an increase in ischaemic segments (+48%, P<0.001) was imaged mainly at the expense of viable (but non-ischaemic) and non-viable segments, which were reduced by 60%. GIK affected stress-induced left ventricular underperfusion only marginally (GIK: 39.7+/-2.5 compared with saline: 35.4+/-2.2 units, P<0.05), but significantly improved its acute reversibility (-42+/-4 compared with -25+/-4%, P<0.001). We conclude that GIK pretreatment attenuates the effect of ischaemia on myocardial contractility, slightly improves exercise tolerance and causes a more rapid and diffuse recovery of post-ischaemic reperfusion.