Objective: We investigated the risk of type 2 diabetes mellitus (T2DM) over a wide range of body iron stores. Methods: Prospective cohort of 1613 men in the Kuopio Ischemic Heart Disease Risk Factor study, aged 42-60 years, free of T2DM and hereditary hemochromatosis at baseline in [1984][1985][1986][1987][1988][1989]. Baseline serum ferritin (sF) and serum-soluble transferrin receptor (sTfR) concentrations were used to predict incident T2DM. T2DM was assessed by questionnaires, blood glucose measurements, and medication reimbursement register. Results: There were 331 cases of incident T2DM during the mean follow-up of 16.8 years (27 098 person-years). At baseline, subjects who later developed T2DM had average sF concentrations of 191 mg/l (S.D. 155) vs 151 mg/l (S.D. 119) among those who remained healthy, P!0.001. In a multivariate-adjusted logistic regression, each 100 mg/l increase in sF corresponded to an average of 14% increased (odds ratioZ1.14, 95% CI 1.03-1.26, PZ0.009) risk of developing T2DM. In a Cox regression, a markedly increased risk of developing T2DM was observed from the fourth sF quintile (185 mg/l, the median) upward (hazard ratio (HR) first vs fifth quintileZ1.5, 95% CI 1.0-2.2, P-trendZ0.05). In a corresponding Cox model in sTfR, the subjects in the third quintile (1840 mg/l, the median) had the least risk (HRZ0.63, 95% CI 0.42-0.97, PZ0.04). Conclusions: Body iron within the sF reference range is not an important determinant of T2DM risk, whereas high normal and above is associated with markedly increased risk. Iron depletion toward iron deficiency as assessed by sTfR is not protective against T2DM. A rule of thumb safe range could be 30-200 mg/l of sF.