Individuals exhibit hindsight bias when they are unable to recall their original responses to novel questions after correct answers are provided to them. Here, I present a series of studies that investigates factors that reduce or eliminate hindsight bias. Specifically, I test the retrieval-based theory, which proposes that hindsight bias can be avoided under task conditions that support the generation of sufficiently discriminative retrieval cues. These discriminative retrieval cues allow participants to selectively retrieve their original judgments, even after being provided with the correct answers. Experiment 1 used the standard memory-design hindsight bias task and a modified design where participants were asked to recall their original judgments and the correct answers. Unexpectedly, participants in the standard memory-design avoided hindsight bias. As predicted, and consistent with the retrieval based theory, participants who engaged in the modified task were also able avoid hindsight bias and were able to recall the correct answers. In order to better understand Experiment 1’s surprising results, Experiment 2 used a think-aloud methodology to determine which retrieval strategies participants were using when they correctly recalled their original judgment. Participants were observed successfully using the discrimination strategy when they were not prompted to do so (standard design), and when they were prompted to do so (modified design), providing support for the discriminative retrieval cue mechanism. Experiment 3 investigated whether sufficiently discriminative retrieval cues continued to reduce hindsight bias after a one-month delay, and whether repeated retrieval contributes to this reduction. Participants were observed engaging in hindsight bias after a one-month delay, suggesting the utility of the discriminative retrieval cues deteriorates over time. Further, repeated retrieval did not mitigate the presence of hindsight bias following this time delay. Understanding the factors that reduce or even eliminate hindsight bias is important because it informs competing cognitive theories of this effect, and because it potentially informs the design of science instruction that minimizes hindsight bias and supports more normative reasoning.