The purpose of this study was to identify the working memory capacity (M-space) for a group of 139 children (4 to 8 years of age) enrolled at Sandy Bay Infant School in Hobart, Tasmania, Australia. Four M-space tests -- The Counting Span Test (Case and Kurland, 1978), Mr. Cucui Test (DeAvila and Havassy, 1974), the Digit Placement Test (Case, 1972), and The Backward Digit Span Test (Hiebert, 1979) -- were administered to the children. Using four different scoring rules for each test, it was found that the context of test may give students a cue which helps them answer questions above their M-space level, that there is no abrupt shift or change in M-space, and that some children appear to incorporate a strategy of chunking information in order to solve problems beyond their indicated M-space level. Cross tabulation of the four tests indicated that children classified into an M-space level by one test would not necessarily be classified into the same level using another test. Analysis revealed one primary factor reflecting quantitative skills and another reflecting spatial perception. The principal conclusion of this study is that with some qualifications children can be grouped by test scores into well-defined sets with similar working memory capacities. (Author/MP)