Since birth, multisensory cues improve our interaction with the environment. For instance, the perceived hum of wings, the vibrations on our skin, and its visual appearance can help us to detect the location of an insect in our body, as well as allow us to guide the motion of our hand to shoo it away. Multisensory processing might occur at different levels: we found that combining tactile and visual stimuli (summation condition) led to a general enhancement in perception, while a different technique (facilitation) revealed direction specific improvements in perception, indicating early sensory interactions. Moreover, in certain circumstances, multisensory integration may not be achieved until the age of 10. Before this age, the noise in precision during active movements increased by double compared to passive ones. To overcome this, we propose that, before multisensory integration occurs, a cross-modal calibration process in which the most reliable sense calibrates the others is necessary. Evidence are obtained by exploring the impact of a lack of visual calibration, essential modality in space representation, on the tactile perception in passive and active conditions. By testing blind and sighted individuals, we found a diminished tactile reliability in the former group during active touch. When auditory stimuli were combined with tactile ones, while blind participants were only affected by the movement, the active touch noise made sighted participants vulnerable to the interference of the sound. These findings support the role of cross-modal calibration as pre-requisite for an efficient multisensory processing as well as in shaping our perceptions.