Research has found evidence for audiovisual integration in several fields related to human social interactions, including speech, identity recognition, and emotion recognition. Surprisingly, few studies have examined audiovisual integration of attractiveness judgments. Are judgements based on sound (i.e., voices) and judgements based on visual information (i.e., faces) combined when assessing overall person attractiveness? If so, what is the predictive value of each modality? We defined several possible outcomes: a) one modality is the better predictor, b) overall attractiveness results from an integrative process of both modalities, or c) overall attractiveness cannot be modelled by any of the modalities. To address this question, we conducted an online experiment (Study 1) using standardized video material recorded under controlled conditions; and a field study (Study 2) using participant-created video material and real-life interactions. In both studies, participants first judged person attractiveness based on information from either audio recordings or muted video recordings, and second based on information from an audiovisual video (i.e., voice and face) in an online experiment. Additionally, Study 2 included a speed-dating event where participants met a subset of the individuals they previously rated. Here they interacted with each other for 4 minutes and judged person attractiveness again, using all information available from a real-life interaction. Together, these studies clarify the relative contributions of faces and voices to person attractiveness judgements and, furthermore, allow us to compare results from online experiences with real life encounters. First results are presented.