In this paper, a new speed sensorless control algorithm is introduced for induction machine (IM) drives, which utilizes the cubature Kalman filter (CKF) as state estimator. The CKF is a relative new approach to nonlinear state estimation, which uses a spherical-radial cubature rule in order to compute numerically the multivariate moment integrals. Using an augmented state space model of the IMs, CKF provides load torque estimation and sufficient control performance under difficult operating conditions. The proposed control algorithm is validated by MATLAB simulation in this work.