Biodiesel fuels are classified into three generations according to their feedstock origin. Biodiesel cost and fuel properties substantially depend on the feedstock type. In this study, to characterize the biodiesel feedstocks of three generations, some critical fuel properties (kinematic viscosity, density, acid, iodine, calorific, peroxide, and saponification values) and also fatty acid compositions of 25 different biodiesel feedstocks (vegetable oils, waste animal fats, waste cooking oils and algae oil) were determined. After the characterization step, statistical studies (correlation-regression analyses) were performed and empirical formulas were developed to estimate the related fuel properties of biodiesel feedstocks from their fatty acid compositions. The investigated biodiesel feedstocks' fuel properties and fatty acid compositions were generally very different from each other, indicating their heterogeneous nature. Overall correlation coefficients obtained for all feedstocks were quite low. Similarly, the general empirical formulas developed for 25 feedstocks did not give very accurate results (except for the fuel properties of waste animal fats). Therefore, statistical analyses were repeated specifically for vegetable oils and algae oil (9 feedstocks). As a result, higher correlation coefficients were determined. Moreover, the fuel properties were estimated much more precisely with the new empirical formulas. The mean absolute relative error obtained using new models were 0.003% for saponification value, 0.027 % for iodine value, 0.074 % for kinematic viscosity, 0.142 % for peroxide value, 4.393 % for acid value, while density and calorific value were calculated without error. The findings could be useful in further academic investigations and also in biodiesel industry for feedstock characterization and waste management.