An emerging body of studies demonstrates that practicing retrieval of studied information, by comparison with restudying or no treatment, can facilitate subsequent learning and retrieval of new information, a phenomenon termed the 'forward testing effect' (FTE) or 'test-potentiated new learning." Several theoretical explanations have been proposed to account for the FTE. A release-from-PI theory proposes that interpolated testing induces context changes and enhances event segregation, which in turn protect new learning from proactive interference (PI). A strategy-change view hypothesizes that prior tests teach learners to adopt more effective/elaborative learning and retrieval strategies in subsequent study and test phases. Finally, a reset-of-encoding account proposes that interim testing on studied information reduces memory load, resets the subsequent encoding process, and enhances encoding of new information. The current study recruited a large sample (over 1,000 participants) and employed a multilist learning task and mediation analyses to test these theories. The results suggest that prior list intrusions (an index of PI) significantly mediated the FTE, supporting the release-from-PI theory. In addition, interim testing enhanced strategic processing of temporal information during new learning (reflected by increased clustering), and temporal clustering significantly mediated the FTE, supporting a role for strategy-change in the FTE. Lastly, a variety of indices were constructed to represent the benefit of reset-of-encoding, but none of them provided evidence supporting the reset-of-encoding view. The results shed new light on the complex mechanisms underlying the forward benefits of testing.