Retrieval practice improves long-term retention. However, it is currently debated if this testing effect can be further enhanced by overtly producing recalled responses. We addressed this issue using a standard cued-recall testing-effect paradigm with verb–noun action phrases (e.g., water the plant) to prompt motor actions as a specifically powerful response format of recall. We then tested whether motorically performing the recalled verb targets (e.g., ?–the plant) during an initial recall test (enacted retrieval) led to better long-term retention than silently retrieving them (covert retrieval) or restudying the complete verb–noun phrases (restudy). The results demonstrated a direct testing effect, in that long-term retention was enhanced for covert retrieval practice compared to restudy practice. Critically, enactment during retrieval further improved long-term retention beyond the effect of covert memory retrieval, both in a congruent noun-cued recall test after 1 week (Experiment 1) and in an incongruent verb-cued recall test of nouns after 2 weeks (Experiment 2). This finding suggests that successful memory retrieval and ensuing enactment contribute to future memory performance in parts via different mechanisms.