It is still an open question whether vision plays a dominant role in space perception. Here, we tested spatial acuity in congenitally blind and sighted volunteers for different unisensory and multisensory spatial coding conditions. In the egocentric condition, participants indicated whether a visual (LED), a kinesthetic (passive right-hand movement), or a visuo-kinesthetic trajectory was left- or right-oriented in reference to the body midline axis. In the allocentric condition, participants judged whether intersecting trajectory segments described either an acute or an obtuse angle. A psychometric function was fitted to the data to define the bias (measure of accuracy) and the uncertainty range (measure of precision). Space perception of sighted participants was more accurate for the combined visuo-kinesthetic information than for the unisensory visual or kinesthetic information suggesting that both vision and kinesthesia contribute to space perception. Sighted participants’ estimates based on kinesthetic input were more accurate than those of congenitally blind participants, irrespective of the spatial coding condition. However, early spatial training of congenitally blind adults improved spatial accuracy and precision matching the performance level of the sighted. This effect was more pronounced for allocentric than egocentric coding. Thus, early non-visual experience of space seems to compensate for the lack of vision.