In this study, the hydromechanical and general efficiencies have been examined experimentally and theoretically by changing the revolution and outlet pressure of pump affecting the efficiencies of hydrostatic pumps such as gear, vane and axial piston pump. This paper discusses a new modelling scheme known as artificial neural networks. The hydromechanical and general efficiencies are predicted using feed forward architecture of neurons. The inputs to the networks are the collection of experimental data. These data are used to train the network using the Batch Back-prop, Online Back-prop, and Quickprop and Delta-Bar-Delta algorithms. The neural network model outperforms the available experimental model in predicting the efficiencies.