Automobile drivers/passengers perceive automatic transmission (AT) shift quality through the torque transferred by the transmission. Clearly, torque regulation is important for transmission control. Unfortunately, a physical torque sensor has been too costly for production applications. With no torque measurement for feedback, controls in AT is mainly implemented in an open-loop fashion. Therefore, complicated adaptation algorithms are necessary while undesired shifts may still occur. To further simplify the controls and enhance its consistency and robustness, a direct torque feedback has long been desired in transmission control synthesis and development.A “virtual” torque sensor (VTS) algorithm has recently been developed to show a good potential in estimating relative torque along transmission output shaft using transmission output speed sensor and wheel speed sensors. This concept has been validated on production vehicles with a good accuracy and robustness in torque estimation under all kinds of driving conditions. This promising technology has since received much attention and is considered as a potential low-cost solution to transmission torque measurement. However, the fact that the VTS estimate is only on the relative torque can greatly limited its use in the transmission control areas.This paper extends the previous VTS work to obtain the absolute torque information based on the framework of speed sensor based torque estimation. We propose a teeth matching method involving hardware and software modifications. An additional speed sensor and a toothed wheel are added on the transmission propeller shaft before the differential. Torque estimation is calculated based on the signals from the transmission output speed sensor and the newly added sensor. The toothed wheels associated with the both speed sensors are machined in such a way that there is a special tooth that can be distinguished from the other teeth. Such a tooth serves as a positional identifier to the shaft. With the two special teeth aligned initially when amounted, any misalignment between them at the presence of load reveals the torsion in the shaft, which is an indicator of the absolute torque being transferred. This absolute torque value can be used as a reference in the VTS algorithm developed previously. The whole concept has been validated on a production vehicle and the estimated torque attained is benchmarked with the torque measured by a physical torque sensor.