Starting from 2000, the Programme on International Student Assessment (PISA) is held by the Organization for Economic Co-operation and Development (OECD) for the 15-year-old every three years. The sample includes over 270,000 participants of 41 countries in the PISA 2003 database aimed at evaluating mathematics and problem-solving ability. Most of the studies on the database are analyses of descriptive statistics, however, the database has the characteristics of hierarchical properties. It is necessary to use hierarchical linear model (HLM) of two levels to analyze the relation between the variables of countries and students. The unit in level-1 is students. The variables belonging to the level-1 are the score of mathematics cognition and the time studying mathematics homework outside the regular mathematics classes. As to the level-2, the unit is countries and the variable is Education Index (EI) from Human Development Reports (HDR) of United Nations Development Programme (UNDP). Also, in considering of the differences between OECD countries and non-OECD countries, the database is divided into two groups. Therefore, the purpose of this study is to investigate the PISA 2003 database by HLM of two levels analysis and to provide some suggestions in education and recommendations for future research.