The incidence of non-alcoholic fatty liver disease (NAFLD) is increasing in Western and Asian countries, including Japan. NAFLD includes the condition of non-alcoholic steatohepatitis, which can progress to end-stage liver disease. Weight reduction based on basal energy expenditure (BEE) is considered to be the only established treatment for patients with NAFLD. However, a formula that is suitable for predicting BEE in Japanese patients with NAFLD remains to be determined. We enrolled 77 Japanese patients who were diagnosed with NAFLD according to histological findings. Their BEE was measured (mBEE) by indirect calorimetry. Physical findings, laboratory data and their predicted BEE (pBEE) values were compared with the mBEE values. All pBEE values were evaluated as a root mean squared error (RMSE) and an accurate estimation. The mBEE values correlated with the patient's weight, skeletal muscle mass, and age. Most of predictive formulae overestimated BEE in NAFLD patients in the present study. In contrast, the Kyoto equation provided an accurate prediction. Most prediction formulae included body weight as a reference of the skeletal muscle mass and were established using data from a healthy study population. However, differences in muscle mass exist among different races, and body composition differs between healthy individuals and those with high BMIs. The improved accuracy of the Kyoto equation is likely due to the similar backgrounds of the patients in the present study. The Kyoto equation is the most suitable formula for estimating BEE in Japanese patients with NAFLD.