The objective of this research was to obtain a pedotransfer function to estimate the carbon content over humid grasslands of Federal District of Brazil. The estimation of carbon contents from low time-consuming data acquisition variables such as soil texture, represented a useful measurement not only to quantify carbon content but also to facilitate calculation of soil carbon storage which depends on availability of data regarding carbon content (grams of C/kg of soil) and bulk density. The pedofunction was described by a nonlinear regression and the evaluation of its efficiency of prediction was obtained based on coefficient of determination (R2), standard error of estimation and graphical interpretation. The carbon contents in Gleysols and Plintosols under humid grasslands were estimated successfully, with coefficient of determination higher than 0.8.