Causal modeling procedures were used to test causal inferences about hypothesized relationships among the major influences on science learning outcomes for a secondary-two (equivalent to grade 8) sample of Singaporean students. A sample of approximately 900 was selected from schools that represented the population of secondary-two students. Instruments were developed to measure attitudes toward science, perceptions of the science teacher, the classroom environment, the home environment, student motivation, and peer influence. Students also took an ability test and a science achievement test. Other variables explored included student use of time and the amount and quality of homework in science. The research design involved the LISREL approach to causal modeling. Attitudes toward science influenced achievement, and the converse was not true. Motivation, classroom environment, and attitudes toward science were the more significant and alterable variables for science achievement. One implication of the results is that science teachers cannot afford to overlook student attitudes. Seven tables present analysis results, and two figures illustrate the discussion. (SLD)