This study aims to analyze and evaluate the implementation of a deep learning approach in Arabic language instruction at MAN 3 Sleman, focusing on lesson planning, instructional implementation, learning assessment, and the challenges encountered by teachers. This study employed a qualitative approach with an evaluative case study design. Data were collected through in-depth interviews, classroom observations, and document analysis involving the school principal, vice principal for curriculum, Arabic language teachers, and Grade XII-D students. Data were analyzed using data reduction, data display, and conclusion drawing, while trustworthiness was ensured through source and methodological triangulation. The findings reveal that deep learning principles were substantially integrated into Arabic language instruction. Mindful learning was implemented by activating students’ prior knowledge, monitoring their learning progress, encouraging questioning, and facilitating reflection through self- and peer-assessment. Meaningful learning was achieved by connecting Arabic vocabulary and grammar with authentic classroom objects, daily life contexts, sentence construction, and dialogue activities that enabled students to apply language in real situations. Joyful learning was fostered through interactive teaching strategies, simple games, visual media, spontaneous quizzes, and a supportive classroom atmosphere that promoted active participation and motivation. Learning assessment integrated diagnostic, formative, and summative assessments to evaluate cognitive, affective, and psychomotor domains. However, several challenges remain, including limited variation in instructional technology, teachers’ incomplete understanding of the deep learning approach, and the continued use of standardized MGMP lesson plan formats. The study recommends strengthening teacher professional development and improving instructional design to optimize the implementation of deep learning in Arabic language education.