Prior studies have consistently found a surprising inaccuracy of people's neuroticism judgments at zero acquaintance. Based on the Realistic Accuracy Model (Funder, 1995), we hypothesize that this is due to a lack of relevance of the situation in which targets are typically observed. Fifty participants were videotaped in a highly trait-relevant (i.e., socially stressful) situation as well as three less relevant situations. An aggregate of self-reports and informant reports was used as the accuracy criterion. Four independent groups of unacquainted observers judged participants' neuroticism based on these short video sequences. Results showed that neuroticism judgments were significantly more accurate for the most trait-relevant situation compared with the other three situations. This finding can be explained using lens model analyses: Only in the most relevant situation did neuroticism predict both visual nervousness and vocal nervousness, both of which in turn predicted neuroticism judgments by lay observers. Our findings show that strangers are sensitive to interindividual differences in neuroticism as long as targets are observed in a trait-relevant situation.