A study investigated the use of multidimensional scaling (MDS), a statistical technique, to examine the underlying structure of an ability test at the item level. Subjects were over 8,500 Egyptian students in the Department of Public Service at the American University in Cairo. The testing instrument used was a 36-item English language test based on an English textbook used in the program. Results suggest that the cloze items, as contrasted with the elision items, are most representative of the total test score, which could be due to the relative familiarity of the item formats or to the formats themselves. Additional explanations for the results are examined. The complexity of the MDS procedure is necessary to reflect the intricacies of the cognitive processes that item-level analysis can reveal. The major drawback, however, is the present inability of specialists in the field to interpret multidimensional findings more meaningfully. Further research is recommended. (MSE)