Oral acarbose, a competitive inhibitor of alpha-glucosidase, has been shown to be effective in decreasing the postprandial rise in blood glucose and insulin. A double blind, cross-over, placebo controlled, randomized study in poorly controlled, non-insulin dependent diabetic patients under treatment with sulfonylureas was carried out. The patients continued receiving sulfonylureas throughout the study period and were randomly allocated into two sequences. In sequence A they received 100 mg tablets tid during 12 weeks; placebo tid during 2 weeks (wash-out period) and finally they were crossed over to placebo tid during 12 weeks. In sequence B, they received placebo tid 12 weeks, placebo tid during 2 weeks and finally acarbose 100 mg tid during 12 weeks. Sixteen patients were included in each sequence; three were excluded from sequence A, one because of side effects, one because of severe neuropathy and one because of change of address. One was excluded from sequence B because of failure to take the sulfonylurea. A slight but statistically significant decrease in weight was observed with acarbose as compared with placebo in both sequences. Significant reductions in postprandial glucose were observed in both sequences with acarbose. Significant reductions in fasting blood glucose were also observed in some visits. Although lower mean values of triglycerides and HbA1c were observed with acarbose, they were not statistically significant. Acarbose had side effects almost in all patients, but decreased on continued therapy. Only one patient had to be excluded for this cause. Acarbose is a useful therapeutic resource in poorly controlled non-insulin dependent diabetic patients in combination with sulfonylureas.