The present study investigated the effects of intensive recasting on second language learners' use of the Japanese aspectual form -te i-(ru) using a time-series design. Four college classroom learners participated in 8 conversational sessions, with the researcher providing recasts during the middle 4 sessions, and 2 of the learners also participated in a delayed-posttest session after 7 weeks. Overall accuracy increased significantly in correlation with the number of recasts provided during the treatment period, and the accuracy rate was retained. However, in contrast to the prediction based on the aspect hypothesis, the progressive use of -te i-(ru) was less accurate than the resultative use, which showed constant increase in every learner's performance. Possible accounts for the results include the role of input in classroom instruction, learners developmental readiness, and complexity of form-function mapping.