Sixteen non-insulin-dependent diabetic patients, mean age 60 years (range 47-69 years) and duration of diabetes 9 years (2-20 years), completed a randomized cross-over study of three 6-week periods separated by 2-week intervals to minimize carry-over effects, in which their usual bread was replaced by either control bread, guar bread (100 g guar/kg wheat flour), or control bread plus a guar granulate. The mean (+/- SEM) intake of guar taken in bread was 7.6 +/- 0.7 g/day (range 3.1-14.3 g/day). The granulate was taken in a dose of 5 g twice daily which provided 8.3 g guar/day. Significant reductions were found in glycosylated haemoglobin after guar bread (11.5 +/- 0.8% to 10.7 +/- 0.8%; p less than 0.02) and after guar granulate (11.2 +/- 0.8% to 10.6 +/- 0.7%; p less than 0.05) compared with control bread. Total cholesterol was also reduced significantly after both guar bread and guar granulate (p less than 0.01, p less than 0.02), the changes being due to LDL-cholesterol. Dietary intakes and body weight did not change. No significant side-effects were reported; 14 subjects found guar bread more palatable than guar granulate but 8 preferred the granulate for its convenience. We conclude that a lower than convential dose of guar can be effective and palatable. The incorporation of guar into food increases its metabolic benefits and palatability.