Arabic Language Education Study Program students must be able to write theses in Arabic. A common obstacle students face is relying on Google Translate to help them translate from Indonesian to Arabic. However, even though it is easy to use, Google Translate still has obstacles when measured using Larson's translation quality indicators. This study aims to improve the quality of Google Translate translation results for Arabic theses using an Android-based term dictionary. This study is a type of classroom action research carried out in 2 cycles, each including planning, action, observation, and reflection procedures. Data collection was carried out through an assessment of the translation results of 22 students selected based on the criteria of being in the process of completing their thesis. The study results showed that of the 12 sub-indicators of translation quality, the aspects of equivalence and suitability of the source and target languages were the lowest quality. Using an Android-based term dictionary significantly reduced error scores and improved translation quality compared to Google Translate as a translation machine. However, the dictionary does not match the efficiency or time savings the translation machine provides. Furthermore, researchers are expected to be able to produce translation machines that can accommodate the intricacies and complexities of the Arabic language in the future.