Spatial limitation in visual information processing was examined with dot-in-matrix patterns by using a probe recognition procedure. The independent variables were the number (1-16 dots) and the position of target dots. Subjects were four undergraduate students. The data were analyzed and discussed from three points of view; span of attention, spatial limitation of recognition and visual attention. The followings became clear: First, the span of position recognition was 4.8. Second, “spatial span of attention” was defined as the range of dot positions at which subjects can perceive target dots with 75% or more accuracy. It extended around the fixation point and shrinked with the increase of the number of target dots. Finally, the distribution of spatial attention was estimated for each target dot condition under the assumption that the hit RT at each probe position reflects the amount of attention allocated there. Distributions estimated were cone-shaped, and the height and extent changed with the number of target dots. It was suggested that spatial limitation (i.e. spatial span of attention) in the processing of spatial, positions can be explained by the notion of distribution of spatial attention.