The influence on glycosylated hemoglobin (HbA1) of formal education as compared with self-monitoring of blood glucose (SMBG) was studied in a randomized 18-month trial. All adult type I diabetics in a community were identified. Forty-one of these patients had had diabetes for 20 years or less. Thirty-seven patients were included in the study and finally randomized into four groups. Ten patients received individual formal education followed by SMBG, eight patients were instructed in SMBG without pre-education, nine patients were given only formal education and 10 patients made up a reference group. Education did not improve the mean HbA1 values. SMBG resulted in a decrease by 2% in HbA1, from 12 to 10% (p less than 0.05). The final HbA1 level, however, did not differ significantly between any of the groups. SMBG was accepted by 80% of the patients. The liability to hypoglycemia was about equal in the four groups. It was concluded that SMBG, but not education, improved metabolic control to a certain degree.