The effect on children's test scores of different item formats used in standardized mathematics achievement tests was investigated. In a Utah school, 40 second grade students completed a mathematics computation test using eight different formats derived from five standardized achievement tests. Identical content, taken from the most frequently occurring content in the standardized achievement tests examined, was tested with each format. Differences in test score between types of formats were statistically significant. These results indicate that what a student appears to have mastered in mathematics computation is substantially influenced by the format of the particular test used. Differences in scores between the various formats are not accounted for by the normative scaling of the different tests. Greater attention should be given to the effect of test item format in selecting and administering achievement tests. (Author/PN)