Dietary starch contains rapidly (RAG) and slowly available glucose (SAG). To establish the relationships between the RAG:SAG ratio and postprandial glucose, insulin and hunger, we measured postprandial responses elicited by test meals varying in the RAG:SAG ratio in n 160 healthy adults, each of whom participated in one of four randomised cross-over studies (n 40 each): a pilot trial comparing six chews (RAG:SAG ratio 2·4–42·7) and three studies comparing a test granola (TG1-3, RAG:SAG ratio 4·5–5·2) with a control granola (CG1–3, RAG:SAG ratio 54·8–69·3). Within studies, test meals were matched for fat, protein and available carbohydrate. Blood glucose, serum insulin and subjective hunger were measured for 3 h. Data were subjected to repeated-measures analysis of variance (ANOVA). The relationships between the RAG:SAG ratio and postprandial end points were determined by regression analysis. In the pilot trial, 0–2 h glucose incremental areas under the curve (iAUC0–2; primary end point) varied across the six chews (P = 0·014) with each 50 % reduction in the RAG:SAG ratio reducing relative glucose response by 4·0 %. TGs1-3 elicited significantly lower glucose iAUC0–2 than CGs1–3 by 17, 18 and 17 %, respectively (similar to the 15 % reduction predicted by the pilot trial). The combined means ± sem (n 120) for TC and CG were glucose iAUC0–2, 98 ± 4 v. 118 ± 4 mmol × min/l (P < 0·001), and insulin iAUC0–2, 153 ± 9 v. 184 ± 11 nmol × h/l (P < 0·001), respectively. Neither postprandial hunger nor glucose or hunger increments 2 h after eating differed significantly between TG and CG. We concluded that TGs with RAG:SAG ratios <5·5 predictably reduced glycaemic and insulinaemic responses compared with CGs with RAG:SAG ratios >54. However, compared with CG, TG did not reduce postprandial hunger or delay the return of glucose or hunger to baseline.