A behavioral defense against disease involves detecting sickness cues in others and responding adaptively, such as by avoiding social interactions. While studies have shown that humans can discriminate sickness cues above chance in faces after sickness induction, whether this discrimination affects approach-avoidance behaviors remains uncertain. Here, we investigated how facial sickness cues influence judgments of trustworthiness, serving as a proxy measure for social avoidance. In a prior study, facial photographs were taken of 21 individuals when sick (two hours after an endotoxin injection causing a transient systemic inflammation) and healthy (following placebo injection). In the current study, participants in two separate experiments viewed these paired facial photographs and were asked, in a two-alternative forced-choice paradigm, to identify which face appeared sick (n = 94) or more trustworthy (n = 82). Participants discriminated sick faces significantly above chance (73.1 %), with females (76.0 %) performing significantly better than males (69.3 %). Additionally, sick faces were perceived as significantly less trustworthy, being selected in only 34.9 % of trials. Notably, the higher the sickness discrimination accuracy for a particular face, the less likely that face was to be judged as trustworthy. Moreover, females (30.5 %) were significantly less likely than males (39.5 %) to judge sick faces as the more trustworthy looking. Individual differences in participants’ disease vulnerability, disgust sensitivity, and frequency of sickness, as well as facial stimulus participants’ inflammatory response intensity measured via interleukin-6 blood concentrations, body temperature, and sickness symptoms, did not predict sickness discrimination accuracy or trustworthiness judgments. Together, these findings suggest that visual sickness cues negatively affect trustworthiness judgments, potentially reflecting social avoidant behaviors towards individuals who appear sick. While judgments of facial trustworthiness may be considered a social inference about whether an individual is safe to approach, future research should also include manifest measures of approach-avoidance in response to sickness cues.