This study examined how well cognitive abilities predict high school students' science achievement as measured by traditional content-based tests. Students (n = 1,651) from four high schools in three states were assessed on their science knowledge, reading skill, and reading strategy knowledge. The dependent variable, content-based science achievement, was measured in terms of students' comprehension of a science passage, science course grade, and state science test scores. The cognitive variables reliably predicted all three measures of science achievement, and there were also significant gender differences. Reading skill helped the learner compensate for deficits in science knowledge for most measures of achievement and had a larger effect on achievement scores for higher knowledge than lower knowledge students. Implications for pedagogy and science assessment are discussed. (Contains 9 tables, 1 figure and 2 notes.)