To provide a motor controller capable of operating from the beginning while locus follow-up deviation is zero without making it necessary to control learning at the time of a real operation.

This motor controller is provided with a basic control part 12 for deciding a speed command Vref(i) by multiplying a position deviation Pe(i) calculated by integrating position deviation increment ΔPe(i) by a position loop gain Kp so that the motor position can be moved so as to be matched with a position command, a learning control part 13 for controlling learning in the same way, and a basic control plus FF learning part 14 for adding a speed command obtained by the control of learning as a feedforward signal to the speed command so that the motor position can be moved so as to be matched with the position command. Then, the signals of those three control parts are switched, and a torque command Tref(i) is inputted to a speed control part 16, and a motor 18 is driven through a current control part 17.