This Brief is the third in the Consortium for Policy Research in Education's (CPRE) series reporting on the policy implications of Third International Mathematics and Science Study (TIMSS) data. It summarizes results of recently completed research that explores in greater detail questions raised in initial analyses. Earlier analyses were based largely on observed differences between instructional variables in the United States and in high-achieving TIMSS nations. This benchmarking approach hypothesized that certain practices increase student achievement. Some of the research summarized in this Brief continues this earlier work by exploring additional differences between policy and practice in the United States and in other TIMSS nations. Results do not yield simple lessons, but recent analyses of TIMSS data continue to raise important issues for improvement of mathematics and science instruction. They also provide guidance about how future efforts at international data collection and analysis might clarify differences in policy and practice and their links with student achievement. The study concludes that the TIMSS experience highlights an important lesson for all those who hope to use international comparative data for improvement at the national and local levels: that these data can provide critical tools for shaping reform agendas, but they must always be interpreted with caution and with rigorous and thoughtful attention to contextual factors that affect educational policies and outcomes.