BACKGROUND The purpose of this study is to determine sensitivity and specificity of body mass index (BMI) based on Center for Disease Control 2000 (CDC) percentiles compared to fat mass index (FMI) as an indicator of being really obese in children. Obesity was compared based on these two indexes among children under study. METHODS This cross-sectional study was conducted on 410 primary school girls aged 8-10 years, in the city of Esfahan. Weight and height were measured and BMI was calculated based on weight divided by height squared. Fat mass was measured by body composition analyzer (BCA) and FMI was calculated by dividing fat mass by height squared. FMI at or above the 90(th) percentile and FMI less than 90(th) percentile of reference data were considered as criterion for defining real obesity and normal adiposity, respectively. Receiver operating characteristic (ROC) curve was used to assess the performance of BMI in detecting obesity on the basis of FMI. Furthermore, the rate of agreement between two indices was calculated using Kappa coefficient P number. RESULTS Mean and standard deviation of FMI and BMI in all children were 6 ± 2.1 (kg/m(2)) and 19.4 ± 3 (kg/m(2)), respectively. The area under the ROC curve for obesity was 0.75. The cutoff point, sensitivity, and specificity of BMI to classify children as obese compared to FMI were 21.2 kg/m(2), 79%, and 73%, respectively. In this cutoff point for BMI (21.2 kg/m(2)), the agreement rate between BMI and FMI for determining obesity status was 0.5 (P < 0.001). CONCLUSIONS Our results indicated 79% of children who were recognized as obese based on FMI, were also classified as obese according to BMI. Twenty-seven percent of children, who were non-obese, were identified as obese based on BMI. It appears that FMI compared to BMI is more accurate in determining obesity, but further studies are required.