HbA1c was measured in 3240 healthy non-diabetic adult individuals in a working population. There was no difference in HbA1c between sexes. The distribution of HbA1c was approximately normal with a slight difference between mean and median values at all ages in both sexes. HbA1c increased with deterioration of glucose tolerance and with all the known risk factors for diabetes (age, obesity, family history of diabetes, history of a large newborn delivery); age but not body mass index appeared as a factor influencing HbA1c independently. In women, HbA1c levels rose particularly at the age of menopause but the use of oral contraceptives or oestrogens made no difference. In both sexes, HbA1c was higher in smokers than in non-smokers. No consistent seasonal variation was observed. Haematologic factors had a negligible influence on HbA1c level. HbA1c was more highly correlated with fasting plasma glucose than with 2 h-plasma/glucose (r = 0.20 vs 0.11). In a stepwise multiple regression analysis, age followed by fasting plasma glucose were the only two significant factors associated with the level of HbA1c. These data indicate that HbA1c is influenced only by factors closely linked to diabetes.