It is often assumed that informant, or peer evaluation is more trustworthy than self-evaluation because the former is less vulnerable to social desirability response style. The current study examines this assumption in three independent samples, conceptually distinguishing between two types of social desirability: desirability due to item characteristics and due to a person's trait characteristics. We found that self-ratings (Studies 1–3) and peer ratings (Studies 2–3) are equally liable to item desirability in two cultures (Canada and China) with the relation further moderated by rater's trait desirability. Results challenge the popular assumption that informant ratings are impervious to social desirability. Relationship closeness has a moderating role, with closer targets rated more favorably than more distant ones. Results demonstrate the importance of conceptually distinguishing between item and trait desirability and are discussed in terms of a motivational account of response inflation.