This study explored the relationship between reading and achievement on a science standardized test. A nonfiction reading subtest and the science section of the Connecticut Academic Performance Test were compared for Grade 10 students at 3 Connecticut high schools. Results showed a moderate-to-strong positive relationship between the variables. The correlation coefficients across the 3 schools for their entire data sets ranged from 0.41 to 0.74. Though this study was not designed to yield conclusions about causation, the consistency of the positive, moderate-to-strong correlations speaks to the existence of a meaningful relationship between the variables. This relationship should cause educators to carefully examine the validity of using language-based standardized tests to assess students' science knowledge, especially for less-proficient readers. (Contains 5 tables.)