People tend to be bad at detecting lies: When explicitly asked to infer whether others tell a lie or the truth, people often do not perform better than chance. However, increasing evidence suggests that implicit veracity measures and potentially physiological measures may mirror observers' telling apart lies from truths after all. Implicit and physiological responses are argued to respond to lies as a threatening stimulus associated with a threat response. Subsequently, people who tell a lie should thus be liked less than those who tell the truth (indirect veracity measure). In terms of physiology, a threat response should be associated with vasoconstriction, which should reduce peripheral cutaneous blood flow, leading to lower finger temperature when confronted with a lie compared to the truth. We test lie detection using explicit and indirect measures, as well as using infrared thermal imaging as a physiological measure of lie detection. Participants (N = 95) observed people telling lies or the truth about their social relationships, during which participants' fingertip temperature was recorded. Results suggested that the accuracy of explicit veracity categorizations remained at chance level. Judgments of story-tellers' likability and trustworthiness as indirect measures of lie detection showed no evidence that observers could tell apart liars and truthtellers: Those believed to be truthellers were liked and trusted more than those believed to be liars, even when this belief was mistaken. Physiological lie detection measured using thermal imaging also failed: Observers' fingertip temperatures, which were expected to show a physiological lie detection response, did not significantly differ between lies and true stories. If at all, the temperature effects pointed in the opposite direction of the lies-as-threat expectations: Fingertip temperatures increased somewhat while confronted with lies compared to true stories.