This study examined the effects of individual student factors and classroom factors on elementary science achievement within and across five countries. The student-level factors included gender, self-confidence in science and home resources. The classroom-level factors included teacher characteristics, instructional variables and classroom composition. Results for the USA and four other countries, Singapore, Japan, Australia and Scotland, were reported. Multilevel effects were examined through Hierarchical Linear Modelling, using the Trends in International Mathematics and Science Study 2003 fourth grade dataset. Overall, the results showed that selected student background characteristics were consistently related to elementary science achievement in countries investigated. At the student level, higher levels of home resources and self-confidence and at the classroom level, higher levels of class mean home resources yielded higher science scores on the TIMSS 2003. In general, teacher and instructional variables were minimally related to science achievement. There was evidence of positive effects of teacher support in the USA and Singapore. The emphasis on science inquiry was positively related to science achievement in Singapore and negatively related in the USA and Australia. Recommendations for practice and policy were discussed.