How do people reason about others when planning deceptive actions? How do detectives infer what suspects did based on the traces their actions left behind? In this work, we explore deception in a setting where agents steal other’s snacks and try to determine the most likely thief. We propose a computational model that combines inverse planning with recursive theory of mind to select misleading actions and reason over evidence arising from such plans. In Experiment 1, we demonstrate that suspects strategically modify their behavior when acting deceptively, aligning with our model’s predictions. Experiment 2 reveals that detectives show increased uncertainty when evaluating potentially deceptive suspects—a finding consistent with our model, though alternative explanations exist. Our results suggest that people are adept at deceptive action planning, but struggle to reason about such plans, pointing to possible limits in recursive theory of mind.