The aim of the study was to elucidate how extracted starches submitted to food processing (or not) can influence plasma insulin and glucose responses in healthy subjects. Native starches from wheat, manihot, smooth peas, or mung beans were tested either raw, as starch gels (boiled and cooled), or cooked and cooled after a preliminary industrial processing: extrusion cooking for wheat, tapioca for manihot, and noodles for mung beans. Eighteen healthy subjects randomly assigned received three different starches under one form of conditioning. All products were submitted to in vitro alpha-amylolysis. Raw manihot starch produced the lowest (p less than 0.05) metabolic responses. Cooking significantly (p less than 0.01) increased plasma responses. However, cooked mung bean noodles gave metabolic responses similar to those of raw products. Close correlations were found between percentages of in vitro starch hydrolysis at 30 min and mean areas under the glycemic curves and the insulinemic curves (r = 0.95, p less than 0.001).