We study the role of face-to-face interaction for gender differences in deceptive behavior and perceived honesty. In the first part, we compare women to men's deceptive behavior using data from an incentivized income-reporting experiment with three treatments. Reporting is fully computerized in a baseline treatment but occurs face-to-face in the second and third treatment. Lies can be detected in the course of an audit, which happens with a given probability in the first and second treatment whereas it depends on perceptions by others in the third treatment. In the computerized baseline treatment, men and women's deceptive behavior is statistically indistinguishable. However, women's truthfulness increases when face-to-face interaction is introduced in the second treatment. In contrast, males' deceptive behavior does not change until the audit probability depends on their perceived honesty in the third treatment. Then, men's truthfulness rises sharply and exceeds women's level of honesty by far. We elaborate on these gender differences in the second part. We conduct an experiment to assess the honesty of videotaped income-reporting statements from a setting identical to the third treatment. Our findings confirm that men anticipate their low perceived honesty, which is consistent with the results from the first part.