In the present work, modeling of ultrasound-assisted intensification of biodiesel production from neem (Azadirachta indica) oil was investigated using four factors three level central composite rotatable design (CCRD) of response surface methodology (RSM). The experimental data obtained through CCRD was used to train the artificial neural network (ANN) model. RSM and ANN models were developed and compared for their predictive and generalization abilities. To evaluate the accuracy of results, additional experiments were conducted which does not belong to experimental design. The results showed that, both models having the ability to predict the experimental data, but ANN was found to be more reliable and superior than RSM. (C) 2014 Elsevier Ltd. All rights reserved.