The development of internet of things and the related sensor technology have been a key driving force for the rapid development of industry and information technology. The requirement of wireless, sustainable and independent operation is becoming increasingly important for sensor networks that currently could include thousands even to millions of sensor nodes with different functionalities. For these purposes, developing technologies of self-powered sensors that can utilize the ambient environmental energy to drive the operation themselves is highly desirable and mandatory. The realization of self-powered sensors generally has two approaches: the first approach is to develop environmental energy harvesting devices for driving the traditional sensors; the other is to develop a new category of sensors-self-powered active sensors-that can actively generate electrical signal itself as a response to a stimulation/triggering from the ambient environment. The recent invention and intensive development of triboelectric nanogenerators (TENGs) as a new technology for mechanical energy harvesting can be utilized as self-powered active mechanical sensors, because the parameters (magnitude, frequency, number of periods, etc.) of the generated electrical signal are directly determined by input mechanical behaviors. In this review paper, we first briefly introduce the fundamentals of TENGs, including the four basic working modes. Then, the most updated progress of developing TENGs as self-powered active sensors is reviewed. TENGs with different working modes and rationally designed structures have been developed as self-powered active sensors for a variety of mechanical motions, including pressure change, physical touching, vibrations, acoustic waves, linear displacement, rotation, tracking of moving objects, and acceleration detection. Through combining the opencircuit voltage and the short-circuit current, the detection of both static and dynamic processes has been enabled. The integration of individual sensor elements into arrays or matrixes helps to