This study compared the effects of reading-oriented tasks and writing-oriented tasks on students' reading comprehension of geometry proof (RCGP). The reading-oriented tasks were designed with reading strategies and the idea of problem posing. The writing-oriented tasks were consistent with usual proof instruction for writing a proof and applying it. Twenty-two classes of ninth-grade students (N = 683), aged 14 to 15 years, and 12 mathematics teachers participated in this quasi-experimental classroom study. While the experimental group was instructed to read and discuss the reading tasks in two 45-minute lessons, the control group was instructed to prove and apply the same propositions. Generalised estimating equation (GEE) method was used to compare the scores of the post-test and the delayed post-test with the pre-test scores as covariates. Results showed that the total scores of the delayed post-test of the experimental group were significantly higher than those of the control group. Furthermore, the scores of the experimental group on all facets of reading comprehension except the application facet were significantly higher than those of the control group for both the post-test and delayed post-test.