Sound waves are one of the important topics studied in physics. However, students' graph representation is still low, leading to their low concept understanding of physics learning. Therefore, it is necessary to optimize the learning process to improve students' graph representation by providing a valid and reliable graph representation test instrument. This study aims to (1) develop a graph representation test instrument on sound wave topic and determine its psychometric characteristics and (2) describe students' graph representation abilities in implementing measurements. The development followed the ADDIE procedure combined with the modified Wilson, Oriondo, and Antonio procedure. The analysis included content validity, item analysis (Rasch model), paired samples t-test, descriptive analysis, and calculation of effect size coefficient. This development produced two packets of graph representation test instrument on the sound wave topic that is considered feasible to use in the test. The results of the measurement implementation show that students have low graph representation before and after learning. This research is expected to provide a new instrument to be implemented in learning and open new insights related to physics learning.