This study aims to examine the improvement of students' written mathematical communication ability in generative learning assisted by teaching aids and discovery learning and describe students' written mathematical communication in terms of self-confidence. The method used in this study is mix-methods. The population in this study were seventh-grade students of Godong Junior High School 1 in the academic year 2018/2019 as many as 5 classes, and samples were selected by cluster purposive sampling technique as much as two classes for the experimental group and the control group. Data collection methods include tests, questionnaires, documentation, and interviews. The results of this study are improving students' written mathematical communication ability with the application of generative learning assisted by teaching aids better than implementing discovery learning. Description of students' mathematical written communication ability with high confidence able to meet all indicators of communication ability written mathematically, with medium confidence is being able to meet several indicators of mathematical written communication ability while low self-confidence can only fulfill the first indicator of mathematical written communication ability.