Background The effects of obesity on prognosis in gastric cancer are controversial. Aims To evaluate the association between body mass index (BMI) and mortality in patients with gastric cancer. Methods A single-institution cohort of 7765 patients with gastric cancer undergoing curative gastrectomy between October 2000 and June 2016 was categorized into six groups based on BMI: underweight (< 18.5 kg/m 2 ), normal (18.5 to < 23 kg/m 2 ), overweight (23 to < 25 kg/m 2 ), mildly obese (25 to < 28 kg/m 2 ), moderately obese (28 to < 30 kg/m 2 ), and severely obese (≥ 30 kg/m 2 ). Hazard ratios (HRs) for overall survival (OS) and disease-specific survival (DSS) were calculated using Cox proportional hazard models. Results We identified 1279 (16.5%) all-cause and 763 (9.8%) disease-specific deaths among 7765 patients over 83.05 months (range 1.02–186.97) median follow-up. In multivariable analyses adjusted for statistically significant clinicopathological characteristics, preoperative BMI was associated with OS in a non-linear pattern. Compared with normal-weight patients, underweight patients had worse OS HR 1.42; 95% confidence interval (CI) 1.15–1.77, whereas overweight (HR 0.84; 95% CI 0.73–0.97), mildly obese (HR 0.77; 95% CI 0.66–0.90), and moderately obese (HR 0.77; 95% CI 0.59–1.01) patients had better OS. DSS exhibited a similar pattern, with lowest mortality in moderately obese patients (HR 0.58; 95% CI 0.39–0.85). Spline analysis showed the lowest all-cause mortality risk at a BMI of 26.67 kg/m 2 . Conclusion In patients undergoing curative gastric cancer surgery, those who were overweight or mildly-to-moderately obese (BMI 23 to < 30 kg/m 2 ) preoperatively had better OS and DSS than normal-weight patients.