Students' ability to write news texts is often hampered by a lack of understanding of the structure and elements of good news texts. This condition causes students' writing results to not meet the criteria for language, completeness of information, and factual accuracy. This study aims to improve students' ability to write news texts through the application of the Two Stay Two Stray (TSTS) learning model in class VII-A students of SMP Negeri 1 Hiliduho. The TSTS model was chosen because it can increase interaction between students, encourage the exchange of information, and foster a deep understanding of the material through structured group discussions. The type of research used is Classroom Action Research (CAR) which is carried out in two cycles. Each cycle includes four stages, namely (1) planning, (2) implementation of actions, (3) observation, and (4) reflection. The research instruments used are news text writing ability tests, observation sheets, and documentation. In cycle I, the analysis results show that the average score of students' writing ability is 45.09%, with the lowest score being 6.25% and the highest score being 93.7%. These results indicate that students' understanding of the structure, elements, and language of news still needs to be improved. After improving the learning strategy in cycle II, there was a significant increase in students' writing skills. The average score increased to 83.19%, with the lowest score being 50% and the highest score being 100%. This improvement indicates that the TSTS learning model is effective in helping students better understand the concept of news writing. Furthermore, the more interactive and collaborative learning atmosphere has been shown to motivate students to actively exchange information and develop ideas. Based on the results of this study, it is recommended that Indonesian language teachers utilize the TSTS learning model as an alternative strategy in teaching news writing, so that the learning process becomes more meaningful and student learning outcomes can be optimally improved.