The purpose of this study is to describe the learning outcomes using a mind mapping-based problem-based learning model to improve the ability to write narratives thematically with SPSS for Windows Version of the Standard 16.0 to find the N-gain value, normality of data distribution, test data homogeneity, and test research hypotheses. The normality test of the initial test of the two classes was normally distributed. The results of the homogeneity test of the initial test of the two classes were homogeneously distributed with sig. = 0.491>0.05, the average similarity test obtained a sig value of (2-tailed) = 0.538 > 0.05, which means that there is no differences before treatment, the test of the difference in the average N-gain of all students (control class with experimental class) obtained the Asymp Value. Sig. (2-tailed) = 0.000, so Sig. (1-tailed) = 0.000//2 = 0.001. 0.001 < 0.05. So it can be concluded that students skill to write narratives thematically who received learning using the problem base learning method based on mind mapping was better than students who received conventional narrative learning as a whole.