abstract the purpose of this study is twofold: on the one hand, it is intended to see what kind of noticing-the –gap activity (teacher generated vs. learner generated) is more efficient in teaching l2 grammar in classroom language learning. on the other hand, it is an attempt to determine which approach of the noticing-the-gap- activity is more effective in the long- term retention of grammar. to do so, the different types of noticing-the-gap-activities were categorized into two categories of teacher-generated and learner-generated activities. then, 120 language learners took a pet proficiency test, and thirty students whose scores were one standard deviation below the mean were randomly chosen and divided in two groups; namely, one experimental and one control group. the experimental group received learner-generated activities and the control group was instructed using the teacher-generated activity for thirty days. the result of the t-test showed that the experimental group outperformed the control group in grammar. for test the long-term retention of the grammar, the same achievement test was repeated after three month of the treatment, the result again proved the superiority of learner-generated noticing-the-gap activity over teacher-generated ones. these have some implications for efl in iranian educational system.