Previous studies have shown that individuals often make inferences based on heuristics using recognition, fluency, or familiarity. In the present study, we propose a new heuristic called familiarity-matching, which predicts that when a decision maker is familiar (or unfamiliar) with an object in a question sentence, s/he will choose the more (or less) familiar object from the two alternatives. We examined inference processes and ecological rationality regarding familiarity-matching through three studies including behavioral experiments and ecological analyses. Results showed that participants often used familiarity-matching in solving difficult binary choice problems, and that familiarity-matching could be applied in an ecologically rational manner in real-world situations. A new perspective on human cognitive processes is discussed in this study.