Some normal people are falsely classified as having impaired glucose tolerance (IGT) if they are given an oral glucose tolerance test (OGTT) when their last meal contained very few carbohydrates. In this study, the duration of carbohydrate restriction was extended to one and three days and the relationship between the carbohydrate restriction and the glucose tolerance after an OGTT was examined. Two different groups of normal subjects were placed on high-carbohydrate (80% carbohydrates) and low-carbohydrate (10%) diets before an OGTT; one group for one day and the other for 3 days. None of the subjects showed impairment of glucose tolerance when placed on the high-carbohydrate regimens. In contrast, 3 of 12 subjects and 2 of 8 subjects placed on the low-carbohydrate diets for 1 and 3 days, respectively, were classified as having IGT. The impairment of glucose tolerance was invariably accompanied by an increase in the fasting plasma free fatty acid level. The longer the period of carbohydrate restriction, the severer was the glucose tolerance impairment. However, the number of subjects who were classified as having IGT did not depend on the duration of carbohydrate restriction. The impairment of glucose tolerance after carbohydrate restriction may be associated with the Randle effect, which is the activation of the glucose-free fatty acid cycle.