ABSTRACT Salbiana Yus. Influence Learning Approach and Mastery of Diction toward Writing Skills (An Experiment in Jakarta 35 Senior High School Students). Thesis. Graduate Program. Universitas Muhammadiyah Prof. DR. HAMKA, 2011. This thesis aims to investigate the effect of learning approach and mastery of diction in student‟s writing skills. In particular, this study aims to obtain data about: (1) The influence of learning approaches to students, (2) mastery of diction, and (3) student‟s writing skill at SMAN 35 Jakarta. The study was conducted at SMA 35 Jakarta. The method whichis used is the experimental method. The population in this study were students at SMA 35 Jakarta, with affordable population consists of students from grade X SMA Negeri 35 Jakarta, totaling 227 students, spread on 6 (six) classes. The sampling technique in this study is taken by using multistage random sampling. That affordable randomized population is taken in simply random in which 76 students will be divided into two classes consisting of 38 students as an experimental group and 38 students as the control group. The analysis of test instruments is using Kuder Richardson formula (KR-20). It produced mastery of diction instrument reliability coefficient of 0.851. With a reliability coefficient of R11 = 0.851. It is concluded that the mastery of diction tests have high reliability. The analysis of test data using test lilliefors normality and homogeneity tests using the Bartlett test. After the meet both requirements, the data were analyzed with two-way analysis of variance (ANOVA) and Tuckey test at significant level x = 0,05. The results of data analysis showed that: First, there are differences in writing skills among students who study with contextual approach and the conventional approach. This can be seen on the test results Qcount = 4.37 > Qtable = 4.26 with significance level α = 0.05. Second, students' writing skills with a mastery of high diction who learn with contextual approach is better than learn with the conventional approach. This can be seen on the test results Qcount = 8.14 > Qtable = 4.26 with significance level α = 0.05. Third, students' writing skills with a mastery of low diction that learning with the conventional approach is better than learning with contextual approach. It can be seen on the results of the test Qcount = 3.53 < Qtable = 4.26 with significance level 0.05. Fourth, there are significant interactions between learning approaches (contextual and conventional) with the mastery of diction (high and low) on student‟s writing skill among the sample. It can be seen on the test results Fcount = 33.77 < Ftable = 4.08 with significance level 0.05.