The role of previous visual experience in relation to spatial representation is investigated by comparing the performance of the congenitally blind, the late-blinded and the sighted blindfolded on a number, of near-space tasks involving mental rotation, mental manipulation and scale transformation, and a far-space task involving the representation of two routes by means of pointing, drawing and the making of spatial inferences. In relation to the former, the role of visual imagery in assisting performance is seriously questioned as accounting for the inferiority of the congenitally blind compared to the late-blinded and the sighted, since scores on a visual imagery test administered to the latter group failed to correlate positively with task performance. However . the poor tactual exploratory strategies observed in the congenitally blind would account for their poor performance on a variety of spatial recognition tasks. In relation to the latter, the congenitally blind tended to perform in a qualitatively different manner to the late-blinded and the sighted blindfolded in all aspects of the task, in addition to performing at a much poorer level than the other groups. Gross errors were found to be due to varying degrees of "egocentric' or 1%se lf-referent' spatial coding strategies inappropriate to such a task. A validation and reliability methodology successfully developed for analysing drawings in the sighted could only be applied to one congenitally blind subject's drawings, the remainder being highly idiosyncratic. The role of previous visual experience in drawing attention to simultaneously existent spatial locations is discussed, and the importance of training the congenitally blind to explore tactual stimuli in a systematic and exhaustive manner, and also to pay attention to external spatial cues is emphasised as being essential for successful mobility and the use of tactual maps.