We measured liver fat content by 3-Tesla magnetic resonance spectroscopy (MRS) in 34 non-to mild obese Japanese subjects with type 2 diabetes, who were not complicated with any liver diseases including clinical fatty liver (liver/spleen ratio of computed tomography [CT] < 0.9) and were not being treated with oral hypoglycemic agents, insulin, or lipid-lowering agents, and analyzed the relationship between liver fat content and body composition and plasma metabolite. The liver fat content is significantly correlated with variables relating to obesity (body mass index [BMI], body weight, fat mass, waist to hip ratio, visceral fat area, subcutaneous fat area, and serum triglyceride), insulin resistance (fasting plasma insulin and homeostasis model assessment of insulin resistance [HOMA-IR]), adipocytokines (serum plasminogen activator inhibitor-1 [PAI-1] and leptin), and serum cholinesterase, but not CT liver/spleen ratio, which is correlated only with fasting plasma glucose, BMI, and HOMA-IR. Multiple regression analysis revealed that the liver fat content is independently associated with serum PAI-1 level (p < 0.001) and BMI (p < 0.05), but not visceral fat area. MRS is a more sensitive method for quantifying liver fat content than CT in type 2 diabetic subjects with nonto mild obesity and without clinical fatty liver. type 2 diabetes; fatty liver; magnetic resonance spectroscopy; plasminogen activator inhibitor-1; body mass index