Vocabulary learning theory suggests that productive recall should strengthen learning of new vocabulary items (Nation, 2001). CALL can provide both the opportunities for productive recall and the feedback to motivate repeated efforts to reproduce new items. The latter capability appears to give CALL some advantages over paper-based exercises, in particular when such learning is done out of class time without immediate teacher feedback. Three experiments are reported. In the first two, a comparison is made of CALL exercises that encourage productive recall with those that encourage passive recall. Initially productive recall appears to give greater learning gains, supporting theory. Over time, the results are comparable. Both groups make substantial gains. Observation suggests that students apply their own strategies to enhance recall, to some extent independently of the exercise type. The advantages and disadvantages of incorporating productive recall into the design of CALL exercises need to be considered together with those of students applying their own strategies to enhance recall. Based on the outcome of this assessment, a sequence of exercises to introduce new vocabulary is designed that encourages passive and active recall as well as other strategies thought to help initial vocabulary learning. In the third experiment, these exercises are delivered over the Web for a whole semester as preparatory materials for in-class work. Outcomes are reported in terms of learning gains but particular attention is paid to evaluating maintenance of student motivation. It is suggested that CALL is an effective way to introduce new vocabulary, that such effectiveness is enhanced by designing exercises based on theory, and that it works well for sustained periods, even in circumstances where student motivation is not necessarily high, when there is close integration with classroom work.