The aim of this study was to determine whether legumes in a physical form which is rapidly digested in vitro give rise to proportionately greater metabolic responses in vivo than legumes which are slowly digested in vitro. Samples of cooked whole and ground lentils were incubated in vitro with pancreatic amylase for 30 min and the percentage starch hydrolysis determined. Grinding the lentils before cooking resulted in a 5-fold increase in the rate of starch hydrolysis (whole lentils 12.1%, ground lentils 60.9%). For the in vivo studies six healthy, young, lean subjects consumed two test meals containing 50 g starch: whole lentils and lentils that had been ground finely before cooking. Postprandial glucose and insulin responses were measured over 4 h. Peak glucose and insulin responses occurred 60 min postprandially for the whole lentils and 30 min postprandially for ground lentils. Although the increase in plasma glucose after ground lentils (1.6 mM) was significantly higher (p less than 0.025) than that after whole lentils (0.9 mM), there was no difference in the magnitude of the insulin responses. These results indicate that, unlike for cereals, the rate of intestinal starch hydrolysis is not the major factor determining the metabolic responses to legumes. By virtue of their low postprandial glucose and insulin responses, irrespective of their physical form and digestibility, legumes would appear to be ideal for inclusion in the diet of diabetics.