The study aimed to examine the relationship between levels of serum eicosapentaenoic acid (EPA), arachidonic acid (AA), as well as EPA/AA ratio and weight loss during hospitalization in participants considered to be overweight, with type 2 diabetes.The study participants included 142 patients who were hospitalized for treatment of type 2 diabetes. We divided the participants into two groups depending on the achievenemt in reduction of bodyweight 3% or more during hospitalization and examined the relationship between serum levels of EPA and AA, as well as ratio of EPA/AA on admission and effectiveness of weight loss under strict dietary therapy during hospitalization, using Cox proportional hazard models.After adjustment was made for several confounders, the hazard ratio of effective weight loss for logarithmical serum EPA was 1.59 (95% CI 1.02-2.49, P = 0.04) and for logarithmical EPA/AA ratio 1.64 (1.03-2.61, P = 0.04), whereas the hazard ratio for effective weight loss for logarithmical serum AA was 1.11 (0.45-2.78, P = 0.82). In addition, after dividing EPA/AA ratio and serum EPA into quartiles based on participant number, the hazard ratio for the highest quartile of EPA/AA ratio was 2.33 (1.14-4.77, P = 0.02), and for the highest quartile of serum EPA 1.60 (0.80-3.19, P = 0.18) compared with the lowest quartile.These results suggest the possibility that EPA is involved in bodyweight change under a caloric-restriction regimen. In addition, EPA/AA ratio was found to be a better predictor of medical intervention for weight loss among overweight patients with type 2 diabetes, compared with serum EPA level.