Non-alcoholic fatty liver disease (NAFLD) is found in one-fourth of the world population and its prevalence is rising. Screening of NAFLD in general population with ultrasonography (USG) is not a cost- effective approach. Alternative methods for prediction of NAFLD are needed. We conducted this study to assess whether basal metabolic index (BMI), waist circumference (WC), waist height ratio (WHtR) is useful for predicting NAFLD. This study was done in a village of Comilla District, Bangladesh from July 2019 to December 2019. Two hundred and nineteen (219) subjects with NAFLD detected on USG and 100 healthy controls without NAFLD were included in this study. The diagnosis of NAFLD was made on the basis of USG by a radiologist. BMI, WC, WHtR were calculated. The area under the curve (AUC) in the receiver operating characteristic (ROC) was calculated to assess the diagnostic ability of BMI, WC, and WHtR for predicting NAFLD. Sensitivity and specificity with optimal cut-off point was calculated. Data were entered into SPSS16for statistical analysis. Mean age in NAFLD group was 45.11±13.50 years. The AUROC values were 0.832, 0.763 and 0.771 for BMI, WC and WHtR respectively. The Cut-off values were 24.4 for BMI, 75.60 for WC and 0.54 for WHtR. Sensitivity and specificity were 80.24%and 79.12% for BMI, 82.34% and 80.45% for WC, 79.56% and 78.98% for WHtR respectively. ROC for BMI, WC and WHtR were above the diagonal line with high sensitivity and specificity. BMI, waist circumference, waist height ratio is useful for predicting NAFLD in rural Bangladeshi population. We can use these simple anthropometric indices as a screening tool for detection of NAFLD in primary health care set up.