Human agency manifests itself as influencing the external world through its own actions. Measuring sense of agency (SoA) is a challenge for research of human agency. Haggard et al. (2002) found that there is a temporal attraction between voluntary action and its sensory consequence. This so-called “intentional binding” (IB) has served as an important tool for quantifying SoA. This paper reviews its research paradigms and cognitive mechanism. In studies of IB, there are two main experimental methods --- Libet’s clock method and interval estimation method. Libet’s clock method measures the timing perception of action and sensory consequence separately when participants report their timing judgment with the aid of viewing a fast-rotating clock. Interval estimation method directly records the subject-reported interval between action and sensory consequence. Some researchers proposed that to use cognitive process to account for IB, it has to answer how CNS establishes causality between voluntary action and sensory consequence. There were two different theories --- predictive inference theory and retrospective inference theory. Predictive inference theory assumes that the CNS predicts the sensory consequence of action and establishes causality if the prediction matches the actual consequence. Retrospective inference theory argues that CNS infers the causality after it receives the sensory feedback. Researchers found that both of them contributed to the timing perception of action in IB. However, there is no evidence for the timing perception of sensory consequence in IB and the details of their joint contribution are not clear. According to the cue integration theory, the timing shift is a result of integration of the action cue and the sensory consequence cue. Final readout of CNS is based on weighted integration of cues where the weight of each cue is inversely related to its reliability. The more reliable the cue is, the more weight it has. Thus, this theory can explain the timing shifts of both events. The paradigm of IB have extensively used in exploring SoA which plays an important role in self-consciousness and social interaction. Future research can use IB to explore the SoA in social condition, some diseases deficit in SoA and the culture difference of SoA. Though cue integration theory can explain IB better, the evidence is not strong enough, especially for the timing perception of sensory consequence. Future studies are warranted to test cue integration and build quantitative model. At present, neural mechanisms underlying IB are poorly understood. Available studies suggest that supplementary motor area is related to IB but a broad network of neural substrates subserve IB. In addition, it’s necessary to refine our understanding of the relation between IB and agency.