Cation exchange capacity (CEC) is one of the most important soil properties. This property can describe many of soil properties such as soil fertility, specific area and soil water content. Whereas measuring this property is expensive, time-consuming and laboratory tools needed, hence, prediction of CEC using pedotransfer function (PTF) and soil easily properties is very important in soil science studies. So, the objective of the present study was to develop regression pedotransfer functions to predict the CEC using fractal dimension of soil particles. Consequently, 106 soil samples of UNSODA dataset were used. Fractal dimension of soil particle size was calculated and then was used to develop a PTF to predict the soil CEC. Performance of suggested fractal regression was compared the existed functions that use other soil properties as input. Results showed that between all soil easily properties only fractal dimension, % clay and organic matter had a significant coefficient. Suggested fractal regression model (R2= 0. 62, RMSE= 5. 3 and ME= 0. 004) and validation (R2= 0. 59, RMSE= 5. 4 and ME= 0. 054) had a better performance that other functions including suggested function, Bell and Vankulen. With considering of good performance of the suggested fractal function, applying fractal dimension that shows effects of soil texture with a number is approved.