A quasi-experimental study investigated the effectiveness of inductive and deductive types of instruction in the acquisition of a complex grammatical structure, the passive voice. The participants were two groups (n = 34) of second-year Japanese students from a vocational school who had three treatment lessons designed for each group embedded into the course curriculum. The quantitative and qualitative data were gathered to seek possible connections between students' preferences and learning gains. Using a pretest and posttest design, we compared the results of recognition and production tests for inductive and deductive instruction groups. Both types of instructions were found beneficial for learners and yielded a significant effect between the production part of the pre- and post-tests, however, the inductive group significantly outperformed the deductive group on the recognition test. The qualitative data indicated students' preferences toward deductive instruction, as it is traditional and well-known in Japan. The results of the study suggest using the inductive type of instruction with simpler grammar and deductive instruction with more complex grammatical structures. The authors advocate the importance of practicing both types of instructions. The study also shows the benefits of using CLT for teaching English in Japan.