The purpose of this study was twofold. First, it aimed to determine whether the application of backward design model has any significant effect on improving the writing ability of Iranian intermediate EFL learners. Second, the study sought to evaluate the superiority of backward design to conventional forward models in the improvement of the writing ability of Iranian EFL learners. The study population was 100 Iranian EFL learners chosen through the administered Quick Placement Test (QPT) in Gooyesh Language Institute, Isfahan, Iran. Based on the results of the test, they were randomly assigned to two samples (50 each); namely, the control and experimental groups. During the treatment, the control group received the writing instruction through a conventional method of forward design, while the participants in the experimental group were taught by the backward design. Both control and experimental samples were homogenized based on learners' responses to one version of IELTS Writing Task 1 used as the pre-test. At the end of the treatment the same task which was used as the pre-test task was also used as the post-test task to measure the improvement of the writing ability of both groups. The results of data analysis showed that backward design teaching approach failed to be more effective than traditional instruction as far as L2 writing ability of Iranian EFL learners was concerned. Furthermore, backward design teaching approach was pedagogically and significantly superior to conventional forward models in the improvement of Iranian EFL learners' writing ability.