A neural network based implementation of space vector modulation of a voltage-sourced inverter has been proposed in this paper. Use of ANN based technique avoids the direct computation of trigonometric function (e. g. sine function) as in conventional space vector modulation implementation. The main disadvantages with conventional implementation are, use of any look-up table implies the need for additional memory, interpolation of non-linear functions lead to poor accuracy and thus to increased harmonics in the PWM waveforms, the use of a look-up table and interpolation demands additional computing time that limits the maximum inverter switching frequency. The proposed scheme is simple and straight forward and avoids the direct computation of non-linear functions. This scheme is evaluated under simulation for a variety of operating conditions of the drive system and comparison is made with the conventional method of implementation. Results show improved performance and robustness of the drive system.