This research aims to investigate (1) whether the use of Content-Based Instruction is more effective than the use of Task-Based Instruction in the teaching of writing skills, (2) whether the use of Content-Based Instruction (CBI) is more effective than the use of conventional method in the teaching of writing skills, (3) whether the use of Task-Based Instruction (TBI) is more effective than the use of conventional method in the teaching of writing skills, and (4) the most effective method among Content-Based Instruction, Task-Based Instruction and conventional method in the teaching of writing skills. This research was Quasi Pretest-Posttest Control Group Design, which use a quantitative method and belonged to an experimental research. The population of the research was all of students in the tenth-grade students of SMK Muhammadiyah Bungoro and used three classes as the sample of research (one control class and two experiment classes). The researcher used cluster random sampling to determine the experiment and control classes. The variables of research consist of two independent variables; they were CBI and TBI, and dependent variable was the students' writing skills. The researcher used a writing test as the research instrument and assigned the students to write a descriptive text to collect the data. In addition, this research used a content validity and did expert judgment to validate the instrument. The researcher also employed inter-rater reliability to measure the reliability of the instrument. To examine the data, the researcher used SPSS v16 then drawn descriptive and inferential analysis as the result of the research. The research findings showed that all of hypotheses were accepted.