The study aimed to examine the development of instruments measuring the ability to speak Arabic. This research uses development-based quantitative methods. The development method used is the development method of Rama B. Radhakrishna. The development process is done through five steps: background, conceptualization, format & data analysis, validity, and Reliability. The research subjects included PBA UMY Study Program students who had at least taken lectures. Determination of construct validity using Exploratory Factor Analysis (EFA) technique, for small-scale tests. Confirmatory Factor Analysis (CFA) techniques for large-scale tests. The results of the readability test show that the measurement model of Arabic speaking ability according to experts in general can be understood and can be implemented in lectures of PBA UMY Study Program. The results of validation of model A and model B instruments are obtained that all items for each aspect have a high validity category. The results of limited trials on the measurement instruments of Arabic speaking ability model A and model B obtained information that all task items had a moderate level of difficulty. The results of large-scale trials show that the instrument has acceptable construct validity and Reliability. The conclusion of the results of this study is based on the results of the validity test and Reliability Statistics, it can be concluded that the results of construct validity and Reliability have qualified as good validity and Reliability.