In this study, biodiesel was produced from waste cooking oil by using sodium hydroxide and methyl alcohol with transesterification method. Three different fuel blends (25, 50 and 75% by volume blending with diesel fuel) were prepared. The densities of fuels were measured at 0.5 degrees C intervals between 0-93 degrees C. The densities of each fuel sample decreased linearly with increasing temperature and diesel concentration. Regression analyses were conducted in MATLAB program and R-2 (coefficients of determination), correlation constants and root mean squared errors were determined. The experimental results were used to train the artificial neural networks. In the present research, a 3-layer back propagation neural network with 15 neurons in the hidden layer was applied. The best R-2 values with mathematical expressions were 0.9996 and 0.9997, respectively. When using artificial neural networks, a R-2 value of 0.9999 was obtained. The comparison of artificial neural network model with different density prediction models showed that the use of artificial neural networks in density prediction is successful.