A recurrent observation in personality judgments is that individuals' ratings of others' personalities are positively linked to their self-description, and that such "assumed similarity" effects appear to be trait-specific. However, the extent of and explanations for assumed similarity have been addressed only insufficiently. To close this gap, we first provide a meta-analytic summary of evidence on assumed similarity of basic personality traits. More importantly, we then critically test different theoretical accounts of assumed similarity (i.e., lack of information, relation to personal values, and known/spurious similarity) in nine studies. Specifically, we investigated assumed similarity of the HEXACO personality traits among strangers, using tailored experimental tests tackling the different theoretical accounts. Across studies, we consistently found the strongest assumed similarity effects for those traits being most strongly linked to personal values: Honesty-Humility and (albeit somewhat weaker) Openness to Experience. For the remaining traits, no consistent evidence for assumed similarity occurred, even when raters had very limited information about the target person. This contradicts that assumed similarity reflects a lack-of-information effect. In turn, the findings could also neither be accounted for by actual similarity, nor by a shared group membership (i.e., spurious similarity) between rater and target. Overall, our studies support the idea that assumed similarity of basic traits is closely tied to personal values and suggest that this finding is attributable to the high personal relevance of value-related traits. This implies that assumed similarity reflects the assumption that others share basic parts of one's identity, even if these others are complete strangers.