Four young adult (18 to 26 years old), nonobese human subjects (two men and two women) with insulin-dependent diabetes mellitus volunteered to consume a series of three diets: baseline (normal daily intake), wheat bran (normal daily intake + 78 gm wheat bran per day), and cellulose (normal daily intake + 30 gm cellulose per day). Wheat bran and cellulose diets both contained 60 gm dietary fiber, with 50% of the dietary fiber from wheat bran or cellulose, respectively. Each patient served as his or her own control. Randomized diets were of 6 weeks' duration, separated by a 4-week "recovery" period. At the conclusion of each diet, subjects were hospitalized and underwent 12 hours of computer-controlled, insulin-glucose infusions. Significant decreases were seen in fasting cholesterol (p less than .05), but the decreases seemed to result largely from the significant reductions in high-density lipoprotein cholesterol. A large reduction in triglycerides was noted with cellulose feeding but not with wheat bran. The mean daily insulin dose decreased (p less than .05) in response to fiber addition (8% and 10% decrease for wheat bran and cellulose feeding, respectively). Mean biostator insulin requirements decreased 11% with wheat bran (p less than .05) but not with cellulose. During biostator monitoring, subjects experienced delayed postprandial blood glucose and insulin-infusion rate peaks with both wheat bran and cellulose feeding. The wheat bran diet reduced peak blood glucose concentration and peak insulin infusion rate in comparison with baseline and cellulose diets. The data suggest that high levels of cellulose or wheat bran are of marginal benefit to insulin-dependent diabetic subjects.