Abstract To compare the variability of glycaemic responses after starchy test meals versus oral glucose (GTT), 10 normal overnight fasted subjects took 75g glucose, 50g carbohydrate from white bread or 50g carbohydrate from a prototype diagnostic oat bar on separate days. Finger-prick blood was taken before and 1 4 , 1 2 , 3 4 , 1, 1 1 2 and 2h after starting to eat. Each test was repeated 3 times by each subject in random order. Coefficients of variation (CV=100×SD/mean) of blood glucose values for repeated tests within subjects were compared by ANOVA. After GTT, 2h blood glucose, 5.4±0.5 mmol/L, was significantly greater than after bread, 4.6±0.2, and oat bar, 4.4±0.2 (p<0.05). After GTT, 7 of the 10 subjects had 2h blood glucose CV >7%, but after bread 7 of 10, and after oat bar 8 of 10 had CV <7%. The mean CV of 2h blood glucose after GTT, 12.9±2.8%, was 2–3 times greater than after bread, 5.2±0.8%, or oat bar, 4.7±0.9% (p<0.01). The results suggest that starchy test meals may allow more precise assessment of carbohydrate tolerance than GTT despite lower glycaemic responses. Further studies in a wider range of subjects are required to confirm this.