This study examined format differences in the measurement of fifth grade students' reading comprehension achievement. Children were given a multiple-choice reading comprehension test, followed 4 weeks later by a constructed response test on 2 of the same text passages. Results indicated that little comprehension of text content was transferred from the multiple-choice to the constructed response test. Significant gender differences were detected in overall test performance and consistency across test formats. Results raise concerns regarding the validity of multiple-choice tests in the measurement of children's reading comprehension proficiency. (Contains 26 references, 4 tables, and 2 figures.) (Author/RS)