The purpose of the study was to investigate the influence of students' background and perceptions on science attitude and achievement. The data analysed came from Booklet 4 given to 17-year-olds during the 1976–1977 National Assessment of Educational Progress (NAEP) survey. Causal modeling procedures were used to analyze the data. In particular, the LISREL method which underlies the LISREL IV computer program, (Jöreskog and Sörbom, 1978) was employed. The influence of five background variables (sex, race, home environment, amount of homework, and parents' education) on three dependent variables (student perception of science instruction, student attitudes, and student achievement) was examined. Sex, race, and the home environment were shown to have substantial influence on student achievement in science. Further, two different models were tested: a model in which attitudes influence achievement and its converse (achievement influences attitudes). The data supported the first model, that is, attitudes influence achievement.