Violations of expectations have been advanced as an explanation for how people make veracity judgments, and previous research has found that unexpected weird behavior is rated as less honest than expected normal behavior. The current experiment (N = 128) varied norms and expectations independently to test four alternative models of veracity judgments. The models included a normative expectation model, an expectancy violation sufficient model, a model based on Expectancy Violation Theory, and a norm violation model. The data were consistent with a norm violation model. Aberrant behavior, whether expected or unexpected, was rated as less honest than normative behavior. Neither expectation violation nor actual message veracity affected deception judgments. These data provide additional evidence of the primacy of behavior over prior expectations in the evaluation of face to face communication.