Starting inrush current and pulsations in the inducedtorque affect the performance of an induction motor.Artificial neural networks (ANNs) and adaptive neurofuzzy inference system (ANFIS) can enhance theperformance of the motor by making a control systemwhich would provide smooth starting to inductionmotor. Dynamic model of induction machine indifferent frames of reference was implemented usingMatlab Simulink. Feed forward back propagation basedand radial basis neural networks were trained, withdata obtained using simulations, to estimatedifferent parameters required by ANFIS to adjustfiring angle of back-to-back connected pairs ofthyristors in AC voltage controller. Inrush currentand pulsations in torque were reduced significantly.Radial basis and feed forward neural networks werecompared for off-line and on-line training, trainingtime, memory required for implementations, number ofneurons, computational procedures and algorithms,reliability of the system and most important cost ofimplementation. Artificial neural networks andAdaptive neuro fuzzy inference system were developedusing tool boxes in Matlab Simulink.