Suppose you were running late to work and had to decide which route to take that would give you the best chance of getting to work on time. How do you come up with the various routes to consider? How do you assess which route will give you the best chance of getting to work on time? In order to make that decision, you may think about all the prior routes you’ve taken and then evaluate each one with some probability of getting the desired outcome. On the surface, the act of generating choices and evaluating their likelihood may seem to have little in common. However, one may be surprised to learn that these processes are closely intertwined. The findings from this project suggest that judgments of likelihood may be constrained by one’s ability to retrieve from semantic memory. In experiment 1, we demonstrate that one’s general ability to retrieve from long-term memory (LTM) may play a critical role in judgments of likelihood and that the nature of the retrieval may relate differentially to different types of event estimation. In experiment 2, we assess different measurement models of memory and find that the type of relation between memory and judgment changes as the function of the type of memory model that one adopts. Finally, combined data across both experiments reveal that how the to-be-judged items are distributed plays a role in judgments and that retrieval ability, specifically, semantic memory, is predictive of probability judgments. Taken together, we argue that the ability to retrieve from LTM plays a critical role in judging the likelihood of an event occurring.