This paper presents a direct field-oriented induction motor drive system with two distinct features: a sliding mode voltage mode flux observer (SMVMFO) and a fuzzy model reference learning controller (FMRLC). The speed of the induction motor is regulated using an FMRLC, which does not require rigorous tuning of the controller's parameters. The rotor flux estimation for the direct field orientation (DFO) is realized by a newly proposed SMVMFO, which is insensitive to the stator resistance variation. The complete drive system is shown to have very good speed tracking performance, accurate flux estimation and field orientation, and high robustness to stator resistance variation, which has been a long-standing problem in the area of motor drives. Extensive simulations are presented for performance evaluation and validation of the proposed control scheme.