Twenty-four nursery school children ranging from 3.11 years to 5.11 years were tested in counting objects that varied in similarity and contiguity. Objects were presented in three arrangements: arrangements composed of subgroups of contiguous identical objects; arrangements without subgroups; and arrangements composed of subgroups of discontiguous identical objects. Significant differences in counting accuracy were found between contiguous arrangements of subgroups and the arrangement without subgroups at the .001 level, between contiguous and discontiguous arrangements of subgroups at the .01 level, and between no subgroups and discontiguous subgroups at the .01 level. The means of correct responses, grouped responses, and percentage of grouped correct responses were higher under the contiguous arrangement of subgroups than under other arrangements. (Author/DT)