In this paper an artificial neural network (ANN) based speed estimator is presented for vector-controlled squirrel cage induction motor (IM) drive. The drive is stable in all operating region and is independent of stator resistance variation. Stator currents, modified stator voltages (Reference values) with stator resistance adaption are used as input to the ANN and rotor speed is treated as the output. For ANN training, Levenberg-Marquardt algorithm is used. Network is first trained for different test data. Finally the algorithm is tested for motoring and regenerating mode considering various loads, speed levels including effect of stator resistance variation. The proposed method is validated through computer simulation using MATLAB/SIMULINK environment.