The significance of high hemoglobin A1 (HbA1) levels (greater than or equal to 8.0%) found in 12.1% of 648 individuals with normal glucose tolerance constituting a part of a representative population sample was examined. Measurement error in HbA1 and/or glucose-tolerance levels was precluded by HbA1 remaining in the same range over 3.5 yr in 89.7% of 29 individuals with initially high and 68.1% of 22 individuals with initially low (less than 6.5%) HbA1. Rate of deterioration to glucose intolerance (6.9%) in the high group during that period resembled the rate (11.8%) in a control group (n = 279). Fasting plasma glucose significantly accounted for only 2.4% of total HbA1-population variance. No correlation of HbA1 was found with other correlates of glucose tolerance or with daily caloric intake and physical activity. A small but significant increment in HbA1 was associated with smoking (7.1 vs. 6.8%, P less than .01) and with clinically overt atherosclerosis (7.3 vs. 6.9%, P less than .01). We conclude that factors unrelated to glucose metabolism are the main determinants of HbA1 level in normal glucose tolerance and play an important role in diabetes as well. These factors have bearing on evaluation of diabetic control by HbA1 and possibly on risk for diabetic complications.