Permanent magnet synchronous motors (PMSM) are highly preferred in electric vehicles due to their high efficiency, good torque-speed characteristics, simple structure and long life. In this paper, the simulation study of direct torque controlled PMSM for electric vehicles was performed in MATLAB by direct torque control (DTC) method using different speed control and speed estimation methods, because output parameters of DTC method such as torque and current, have high ripple, and this ripple should be reduced for electric vehicle. Due to the simple structure of the DTC and its low dependence on the system parameters, it is widely used in alternative current motors. When DTC is used in PMSM drive, it is known that although it has fast dynamic response when it is compared to other vector control methods, the ripple in the torque and flux increases. In this study, direct torque controlled PMSM is tried to be minimized by using different speed control methods and speed estimation methods. For this purpose, proportional-integral (PI), fuzzy logic and sliding mode speed controller are used as speed controllers and their performances are compared. In addition, speed sensor less direct torque control of PMSM was performed by using sliding mode observer and model reference adaptive system (MRAS) observer. As a result of the simulation studies, it was found that sliding mode speed control method provides less ripples in the torque, better speed control and less energy consumption. Furthermore, as a speed observer, the sliding observer gives better results in speed estimation and provides less energy consumption.