This study aims to determine the effect of the learning method (expository and authentic) and the level of confidence in the ability of religious poetry interpretation of the students of the third semester, majoring in the Indonesian Language and Literature Education of Universitas Negeri Manado. The method used is the quasiexperimental method with 2 x 2 factorial designs. The measurement of Y variable (ability to interpret the religious poetry) uses the writing test and the level of confidence uses a questionnaire. Data analysis technique in this study is analysis of variance (ANOVA) followed by two lanes and Tuckey test to look at the interaction of the group. Before the test, the hypothesis is that analysis requirements normality data test using Liliefors test and homogeneity test data using Bartlett test. The results show that the expository method is better than the authentic method to improve the ability of the students. To improve the ability of the students to interpret the religious poetry, it is better to use the expository method for the group that has a high level of confidence. Also, the lower level of confidence used to the authentic method for the group.