Pseudocontingencies are inferences about correlations in the environment that are formed on the basis of statistical regularities like skewed base rates or varying base rates across environmental contexts. Previous research has demonstrated that pseudocontingencies provide a pervasive mechanism of inductive inference in numerous social judgment tasks (Fiedler, Freytag, & Meiser, 2009). The present research extended the analysis of pseudocontingencies from social judgment to actual choice behavior in a decision scenario of personal relevance. In 4 experiments, participants were first exposed to a learning environment in which choice options were presented together with positive or negative outcomes. The base rates of options and outcomes were skewed and varied across different contexts. After the learning phase, participants chose between options on the basis of the previously learned outcome probabilities and were rewarded in accordance with their individual performance. The results revealed that participants inferred a pseudocontingency between options and outcomes and followed the pseudocontingency in their decision behavior (Experiments 1-4). The observed pseudocontingency was stronger in a context with predominantly positive outcomes and replicated with different learning distributions. Pseudocontingency effects were related to interindividual differences in risk aversion and moderated by ease of base rate learning (Experiment 2) and processing time (Experiment 4), whereas the salience of rare events with extreme outcomes did not affect choice behavior (Experiment 3). The findings underline the role of pseudocontingencies in choice behavior as a subjectively cogent tool for decision making in complex probabilistic environments.