Field studies indicate that people may form impressions about potential partners' HIV risk, yet lack insight into what underlies such intuitions. The present study examined which cues may give rise to the perception of riskiness. Towards this end, portrait pictures of persons that are representative of the kinds of images found on social media were evaluated by independent raters on two sets of data: First, sixty visible cues deemed relevant to person perception, and second, perceived HIV risk and trustworthiness, health, and attractiveness. Here, we report correlations between cues and perceived HIV risk, exposing cue-criterion associations that may be used to infer intuitively HIV risk. Second, we trained a multiple cue-based model to forecast perceived HIV risk through cross-validated predictive modelling. Trained models accurately predicted how 'risky' a person was perceived (r = 0.75) in a novel sample of portraits. Findings are discussed with respect to HIV risk stereotypes and implications regarding how to foster effective protective behaviors.