Compared with frequency domain load identification, time domain method is not only applicable to linear systems but also to nonlinear systems. Besides, time domain method can deal with transient impact force effectively. Consequently, more attention has been concentrated on time domain method in recent years. Development surveys of identification approaches widely utilized in time domain are summarized, such as deconvolution method, sum of weighted acceleration technique, function approximation technique, Kalman filter and recursive least square method, inverse system method and burgeoning intelligent algorithms. In addition, the merits and demerits of these approaches are analyzed respectively. With respect to research object, identification method and application area, the authors present that nonlinear system and linear time-variant system will be the focus of study objects, inverse system method combined with modern intelligent algorithm will be new research hotspot, and application of load identification to fault diagnose will be development trend.