© 2015 Leivers et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Mechanisms enabling men to identify women likely to engage in extra-pair copulations (EPCs) would be advantageous in avoiding cuckoldry. Men's judgments of female sexual faithfulness often show high consensus, but accuracy appears poor. We examined whether accuracy of these judgments made to images of women could be improved through i) employing a forced choice task, in which men were asked to select the more faithful of two women and/or ii) providing men with full person images. In Experiment 1, men rated 34 women, for whom we had self-reported EPC behavior, on faithfulness, trustworthiness or attractiveness from either face or full person photographs. They then completed a forced choice task, selecting the more faithful of two woman from 17 pairs of images, each containing one woman who had reported no EPCs and one who had reported two or more EPCs. Men were unable to rate faithfulness with any accuracy, replicating previous findings. However, when asked to choose the more faithful of two women, they performed significantly above chance, although the ability to judge faithfulness at above-chance levels did not generalize to all pairs of women. Although there was no significant difference in accuracy for face and full person image pairs, only judgments from faces were significantly above chance. In Experiment 2, we showed that this accuracy for faces was repeatable in a new sample of men. We also showed that individual variation in accuracy was unrelated to variation in preferences for faithfulness in a long-term partner. Overall, these results show that men's judgments of faithfulness made from faces of unfamiliar women may contain a kernel of truth.