Evaluating other people’s sincerity is a ubiquitous and important part of social interactions. Eleven experiments (total N = 6965; seven preregistered; seven in the main paper, four in the SOM; with American and British members of the public, and French students) show that response speed is an important cue on which people base their sincerity inferences. Specifically, people systematically judged slower (vs. faster) responses as less sincere for a range of scenarios from trivial daily conversations to high stakes situations such as police interrogations. Our findings suggest that this is because response delays are perceived to be the result of the responder suppressing automatic, truthful thoughts. People also seem to have a rich lay theory of response speed, which takes into account a variety of situational factors. For instance, the effect of response speed on perceived sincerity is smaller if the response is socially undesirable, or if it can be attributed to memory effort. Finally, we showed that even when instructed explicitly, people are only partially able to disregard response speed in making sincerity inferences. Our findings not only help ascertain the role of perceived response speed in interpersonal inference making processes, it also carries important practical implication. In particular, the present study highlights the potential biases that may be observed in judicial settings, since response delays of innocent suspects may mislead people to judge them as insincere and hence guilty.