We examined whether fatty liver, as diagnosed with abdominal ultrasonography, is an independent risk factor for diabetes mellitus during 10 years of follow-up. A total of 840 subjects (467 men and 373 women) were followed for the entire 10 years. The criteria for being non-diabetic were having no history of diabetes, having a fasting plasma glucose level of less than 110 mg/dl and a serum hemoglobin A1c level of 6.4% or less. We indicated that every examine received all examinations after 12 hours of fasting. Well-trained technicians performed abdominal ultrasonography. Although univariate analysis revealed that the presence of fatty liver was related to hyperglycemia 10 years later, multiple logistic regression analysis did not support this finding. In the multiple logistic regression analysis fasting plasma glucose levels at the baseline and age were significantly related to hyperglycemia (odds ratio [OR] = 1.16, 95% confidence interval [CI]: 1.11-1.21, OR = 1.07, 95% CI: 1.01-1.14, respectively). Fatty liver was not an independent risk factor for hyperglycemia in our follow-up study 10 years after the first diagnosis. The high fasting plasma glucose levels were a risk factor for diabetes, even in the normal range.