To define the type of dietary fibre of fibre analogue with the greatest potential use in diabetic treatment, groups of four to six volunteers underwent 50-g glucose tolerance tests (GTT) with and without the addition of either guar, pectin, gum tragacanth, methylcellulose, wheat bran, or cholestyramine equivalent to 12 g fibre. The addition of each substance significantly reduced blood glucose concentration at one or more points during the GTT and generally reduced serum insulin concentrations. The greatest flattening of the glucose response was seen with guar, but this effect was abolished when hydrolysed non-viscous guar was used. The reduction in the mean peak rise in blood glucose concentration for each substance correlated positively with its viscosity (r = 0.926; P less than 0.01), as did delay in mouth-to-caecum transit time (r = 0.885; P less than 0.02). Viscous types of dietary fibre are therefore most likely to be therapeutically useful in modifying postprandial hyperglycaemia.