A sensorless speed control method for induction motors in fuel cell vehicle is presented in this paper. An Artificial Neural Network (ANN) estimates the speed, and a Neuro-Fuzzy controller (NFC) is used in speed control loop, to overcome the nonlinearity of the plant. A PI controller controls the motor flux and the NFC determines the required torque. The tuning of NFC is simple and this is one of the advantages of NFCs compared with the conventional PI controllers. In addition, the nonlinear behavior of NFC increases its robustness against variation of parameters in the plant. The speed estimation is down by a two-layer on-line neural network in the rotating coordinate fixed with rotor flux. The ANN estimator has a simple structure, and its parameters are adjusted on-line. The simulation and experimental results are given to prove the effectiveness of this approach.