The present study investigated the effectiveness of inputbased vs. output-based practice in learning an unsalient grammatical aspect (obligatory null objects) in English complex infinitival structures. The participants were 57 Persian learners who were divided into three groups (input-based, output-based and control), received instruction for three weeks and took two types of grammar tests (comprehension and production) immediately after treatment and two weeks later. Results of the immediate post-tests indicated similar effects for both types of practice on improving learners' receptive knowledge while only the output group showed significant gains on the production test. Furthermore, gain maintenance was observed in the performance of the output group on both measures of delayed post-test. The pedagogical implications arising from these findings are also discussed.