This paper presents an adaptive fading extended Kalman filter (AFEKF) based speed-sensorless induction motor (IM) drive. Conventional extended Kalman filters (CEKFs) assume the system (Q) and the measurement (R) noise covariance matrices as constant, but those matrices are affected by the operating conditions of IMs and deteriorate the estimation performance. To eliminate this adverse effect, an AFEKF algorithm which has the ability to update Q and R matrices according to the operating conditions of IM are proposed, and the stator stationary axis components of stator currents, the stator stationary axis components of rotor fluxes, the rotor mechanical speed, and the load torque including viscous friction term are estimated. To illustrate the superiority of AFEKF-based speed-sensorless IM drive, the control performance of the proposed drive system is compared to that of CEKF-based speed-sensorless drive system under simulations. In addition to the comparison results, the computational burdens of AFEKF and CEKF algorithms are also examined.