Preference reversal (PR) reveals that preferences over risky bets can be reversed between choices and willingness-to-accept or -pay. To date, neither has literature in gambling situations paid attention to whether the expected value difference between bet pairs affects the likelihood of PR, almost nor has empirical research shed light on whether episodic memory is involved in PR. In a laboratorybased study, Experiment 1 varied bet pairs in expected value in a market-like scenario. Undergraduates (N = 64) first completed classic dual-procedure choice and price tasks and then performed a memory test for previous choices. Consistent with past work, participants exhibited non-negligible rates of PR between choices and valuations. The results suggest a tendency that the larger the expected value difference between bet pairs, the larger the predicted PR rate, and provide the first evidence that correct retrievals of initial choices can ameliorate PR. In a subsequent Experiment 2, participants (N = 86) were incentivized to complete choice and price tasks and a memory test on purely risky bets in a pictorial form. We found equivocal evidence of the effect of expected value difference within bet pairs on attraction effect PR, no effect of expected value difference or level on correct recollections, and again substantial evidence that correct retrievals of initial choices can ameliorate PR. Overall, these results reaffirm the existence of the traditional and contextual PR phenomenon and provide evidence about how memory retrieval operates as individuals perform binary choice and pricing tasks.