Abstract: ObjectiveTo explore the correlation between body mass index (BMI) and percentage of body fat(BF%) and the inpact of gender and age on the correlation. MetbodsData from 78, 257 participants in 2006 annual health screening were used for data mining, correlation analysis, linear regression and agreement analysis. Results There was a moderate correlation between BMI and BF% (r=0.641, P<0.001).However, close correlations in bothmale and female(r=0.808, r=0.925, P<0.001, respectively) were found.It suggested based on partial correlation coefficient that gender had much more inportant inpact on the correlation coefficient between BMI and BF% than age. The linear regression analysis suggested that the cut-off points of BMI for ovenv eight and obesity were 22 and 24.5 in male, 19.5 and 22 in female aged less than 30; 24 and 27 in male, 21 and 25 in female aged more than 30, respectively. The agreement rate was 59.2% when original diagnostic standard of BMI was used, with a Kappa coefficient of 0.379±0.003(P<0.001).The agreement rate was 73.6% when new diagnostic standard of BMI was used, with a Kappa coefficient of 0.588±0.002(P<0.001).Twice more ovenveightpopulation could be screened out by the new standard than the original one. ConclusionBMI would be a betterm easurement of body fatty ith new cut-off points established for different gender and age.