The purpose of this study was to examine the science achievement outcomes for different subgroups of students using different assessment formats. A nationally representative sample of approximately 3,500 ninth grade science students from 13 high schools throughout the United States completed a series of science assessments designed to measure their level of achievement on the national science education standards. All of the schools were using a curriculum designed to meet the standards. The assessments included a multiple-choice test, a written open-ended test, a hands-on lab skills test, and a hands-on full investigation. The results show that the student outcomes on the different assessment formats are more highly correlated for higher achieving students than for lower achieving students. Patterns for different cultural groups also vary by assessment format. There were no differences found for sex. The results support the notion that different assessment formats assess different competencies and that the achievement of students from different subgroups varies by assessment format.