Above-level testing is the practice administering a test level—of usually an academic achievement or aptitude test—to a gifted or high achieving child. This procedure is widely accepted in gifted education circles, on the basis of theoretical claims that above-level testing raises the test ceiling, increases variability among gifted students’ scores, improves reliability of data, reduces regression toward the mean, and improves interpretation of data from gifted students. However, above-level testing has not been subject to careful psychometric scrutiny. In this study, I examine reliability data, growth trajectories, distributions, and group differences of above-level test scores obtained from the Iowa Tests of Basic Skills (ITBS) and Iowa Tests of Educational Development (ITED). Subjects in this study were 224 students tested a total of 435 times while enrolled in a gifted magnet program for middle schoolers. Longitudinal analyses performed with hierarchical linear modeling indicate that substantial differences exist between students from overrepresented ethnicities (White and Asian Americans) and those from underrepresented ethnicities (Hispanic and African Americans) in both initial scores and the rate of score gains. Gender differences existed only for the rate of score increases for above-level reading scores. Socioeconomic differences existed, but did not have a unique impact beyond that of the ethnicity variable. A discussion of the place of these results within the wider gifted education research context and suggestions for further research are included. An appendix to the study gives information about item difficulty indexes for every item in the ITBS/ITED core battery for the eighth, ninth, and tenth grade levels of Form C.