The purpose of this study was to apply generalizability (G) theory in evaluating the reliability and validity of the test on the performance of the badminton serve. A random sampling model for validity by Kane (1982) was utilized to examine the coefficient validity of the test and the generalizability studies (G-STUDY) by Cronbach (1972) was applied to infer the test's reliability. The subjects were 554 students from NEHS, inc1uding fourth and sixth graders in the elementary department and freshmen to senior students in the senior high department. AII of the subjects took the same test at the end of 3 weeks of course work. The obtained data were analysized by the GENOVA analysis system and the results proved that the design of this study's test had better validity and reliability than past studies and therefore was suited for use with high school students and elementary school students.