The aim of this research is to know empirical evidence whether the English Summative Test for the Third Grade Students of M.Ts. Al ? Gaotsiyah, Jakarta Barat 2009/2010 academic year is reliable or not. The method that the writer uses in this study can be classified as a descriptive quantitative analysis method. The study is based on the research field to obtain research?s supporting data. This research was done in M.Ts. Al ? Gaotsiyah, Jakarta Barat, using of the test re - test method. Collection data was done by using the parallel test as an instrument of research. The writer took the population from the students of the third grade of M.Ts. Al ? Gaotsiyah, Jakarta Barat. There are 47 students from 2 parallel classes. Every class contains about 20 pupils. From all of the students, the writer chooses two classes that contained 40 students as sample. The collected data was analyzed by using Pearson?s Product Moment Correlation. The hypotheses for this analysis are the Hi (research hypotheses) and H0 (null hypotheses). The Hi explains that the English summative test for the third grade students at M.Ts. Al ? Gaotsiyah, Jakarta Barat 2009/2010 academic year is reliable and H0 explains that English summative test for the third grade students at M.Ts. Al ? Gaotsiyah, Jakarta Barat 2009/2010 academic year is unreliable. The result was showed that the index of correlation between English summative test group of scores and the parallel test of English summative test group of scores / ro (0.95) is higher than the critical value of the correlation from the table of standard correlation significance of Pearson?s Product Moment Correlation by df = 38 / rt (1.70) and rh (3.58). Therefore, there is a high significance correlation between the X and Y group of scores. Based on this statement that H0 is rejected and Hi is accepted, and it can be concluded that the scores of students? responses to the test model and the parallel test have been significantly correlation, it has shown the stability.