The purposes of this study were to develop a mathematics achievement test for the seventh graders in Tainan based on the Grade 1-9 Curriculum, and to explore the difference in performance for variables of gender, tutoring, scale of school, public-private school and textbook version. The population of this study consisted of the eight graders in Tainan City. Three samples were drawn for this study: the first tryout sample¡]n=402¡^, the second tryout sample¡]n=455¡^and the norm sample¡]n=870¡^. The Mathematics Achievement Test for Seventh Graders developed by the researcher was used. The data were analyzed in two dimensions: 1. items and test with methods of CTT and IRT; and 2. the difference in performance for variables of gender, tutoring, scale of school, public-private school and textbook version with statistical testing methods. The results of this study were listed below: I. Item and Test Analysis: 1.Regarding the difficulty and discrimination of CTT, the average difficulty of interval scale¡]¡µ¡^of items was 12.14, the mean of percentage passing and discrimination were 0.58 and 0.52, respectively. 2.According to the result of the reliability examination, the Cronbach value of internal consistency was 0.85, the test-retest reliability indicated 0.86, the split-half reliability showed 0.86. 3.The evidences of validity about this test were well content-related, criterion-related, construct-related. As for the result of the validity examination, the content validity of this test was through experts¡¦ review and judgment. The criterion validity was based on significantly positive correlation of the monthly test scores for each school. The construct validity had established in two ways of the internal consistency and the different groups analysis. 4.Analyzing items with IRT, 3PL model, the mean of discrimination (a), difficulty (b) and guessing(c) are 1.23, 0.14 and 0.21, respectively. 5. DIF was present for the item 17 between gender groups. 6. The correlations of item difficulty values between IRT and CTT were 0.97¡ã1.00, the correlations of item discrimination were 0.21¡ã0.95. The correlation between ability parameters of IRT model and total score of CTT ranged from 0.98 to 1.00. The result of indices between IRT and CTT revealed high correlations. 7. This test established a percentile rank and T-score norms. II. Testing of the performance differences with significance level£\=0.01: 1.The student¡¦s performances were significantly different between gender groups, and the female exceeded male. 2.The performances of the students who accepted tutoring after school hours were better than the others¡¦ performances significantly. 3. The student¡¦s performances in the public and private school were significantly different, and the private school outperformed the public school. 4.The performances of the students in private school were better than the performances of the students¡]both accepting tutoring and not ¡^in public school. And in public school, the performances of the students accepting tutoring were better than the students not accepting tutoring. 5.As for the school scale, the large is better than the small, and the middle is better than the small at statistical significant level. 6.In the textbook version, this test found that the performance of student using the version Nani surpassed the performance of student using the version Hanlin, and the version Jenlin outshined the version Hanlin. Finally, based on the results and findings of this study, numerous suggestions for the practical use of the test and future researches were provided.