This article explores the influence of interviewer's race on skin colour classification for white and African American survey respondents using data from the Multi-City Study of Urban Inequality (conducted from 1992 to 1994). The study revealed a compelling race-of-the-interviewer effect for both black and white respondents: White interviewers reported the skin tones of black respondents as substantially darker than did black interviewers, while black interviewers categorised the skin tones of white respondents as much lighter than did white interviewers. Interviewers also perceived greater variation in the skin tones of same-race respondents than among other-race respondents. Study results suggest that both black and white Americans exhibit relatively limited ability to distinguish the physical characteristics of other-race people, and that unsuccessful attempts to match interviewers and respondents by race may have the unintended consequence of introducing important attenuating biases into analyses involving skin colour.