Mortality is increased in type 2 diabetes mellitus (T2DM). Although previous studies showed that decreased serum insulin-like growth factor-I (IGF-I) levels are associated with diabetic complications, little is known about the association between serum IGF-I level and the mortality in patients with T2DM. This is a historical cohort study with end-point of all-cause mortality in 234 men and 191 women with T2DM. Standard deviation of serum IGF-I [IGF-I (SD)] was calculated by adjusting for age and gender. Of 234 male and 191 female, 46 and 25 patients died, respectively, for the follow-up period of almost 7 years. Unadjusted survival analyses showed that lower IGF-I was associated with higher mortality in men and women (p<0.001 and p=0.004, respectively). In Cox regression analyses adjusted for age, duration of diabetes, body mass index, HbA1c, and serum creatinine, serum IGF-I was inversely associated with the mortality [men, hazard ratio (HR)=0.40, 95% confidence interval (CI)=0.25-0.64 per SD increase, p<0.001; women, HR=0.28, 95%CI 0.08-0.96, p=0.043]. When IGF-I was replaced to IGF-I (SD), the relationships are still significant. After additional adjustments for serum albumin, systolic blood pressure, ALT, LDL-cholesterol, smoking, and past history of cardiovascular disease, the association between serum IGF-I and the mortality in men remained significant (HR=0.31, 95%CI 0.15-0.65, p=0.002), but not in women. The present study showed that lower serum IGF-I levels were associated with the increased all-cause mortality in patients with T2DM, suggesting that serum IGF-I could be clinically useful for assessing the risk of mortality in the population.