Two experiments determined whether metamemory judgments invoking covert retrieval practice for a list of unrelated paired associate words led to the facilitation of learning a subsequent list. Three types of relation between successive lists were compared: negative transfer (A-B, A-D); a control for item-specific proactive interference (A-B, C-D); and repetition (A-B, A-B). Experiment 1 showed that the benefit of retrieval practice relative to restudying was equivalent for overt and covert retrieval in the negative transfer paradigm (A-B, A-D). Both types of retrieval minimized intrusions of first list responses in the cued recall of the second list. Experiment 2 showed that memory enhancement following covert retrieval was equivalent for new (C-D) and repeated (A-B) lists. The results are consistent with theories of the forward testing effect (FTE) that assume retrieval practice insulates subsequent learning from proactive interference and provides self-assessment feedback that may lead to more efficient encoding in future learning. Challenges in accounting for the impact of a substantial number of moderators of the FTE are reviewed.