When we actively interact with the environment, it is crucial that we perceive precise temporal relationship between our own actions and subsequent sensory effects to guide body movements. Despite its significance, however, little attention has been paid to the effects of voluntary action on temporal perception of sensory events. The present study examined how voluntary action affects the performance of temporal order judgment (TOJ) for pairs of auditory and somatosensory events. In the voluntary condition, participants actively pressed a button and a noise burst was presented at various onset asynchronies relative to the button press. The participants made either 'sound-first' or 'touch-first' responses. We found that the TOJ performance in the voluntary condition (as indexed by just noticeable difference) was significantly better than that when their finger was passively stimulated (passive condition). We further examined whether the performance improvement with voluntary action can be attributed to the predictability of the timing of the stimulation, the kinesthetic cues contained in the voluntary action, and/or the existence of the motor command. When three noise bursts were presented before the target burst with regular intervals (predictable condition) and when the participant’s finger was moved passively to press the button (involuntary condition), the TOJ performance was slightly better than that in the passive condition, but the improvements failed to reach statistical significance. These results suggest that the improvement in sensitivity to temporal asynchrony between auditory and somatosensory events caused by the voluntary action cannot be fully attributed to sensory predictions and kinesthetic cues. Rather, the existence of the motor command and its efference copy would be crucial for improving temporal sensitivity.