Studies indicate an ethnic density effect, whereby an increasing proportion of persons of the same ethnicity as oneself (co-ethnics) in one's area of residence is associated with reduced risk of morbidity among ethnic minorities, though evidence is mixed. Measures of ethnic density are commonly taken from small-area census data using predefined categories of ethnicity. In a United Kingdom study, the authors compared these measures with perceived ethnic density, based on self-reported proportion of co-ethnics in the area. Using 2005 Home Office Citizenship Survey data linked to the 2001 United Kingdom Census, they found moderate-sized correlations between perceived and measured ethnic density which varied across ethnic groups (r = 0.34-0.65). Perceived ethnic density underestimated measured levels for whites and overestimated measured levels for ethnic minorities. Compared with participants in areas where less than half of residents were co-ethnics, those reporting a perceived ethnic density of more than half tended to have less limiting long-term illness (for all ethnic minorities combined, odds ratio = 0.81, 95% confidence interval: 0.63, 1.04) after adjustment for age, sex, socioeconomic position, ethnicity, area deprivation, and measured ethnic density. After adjustment for perceived ethnic density, there was no evidence of a protective association for measured ethnic density, except for Caribbeans. Perceived ethnic density may reflect individual experiences of frequency and intensity of contact with co-ethnics, which may explain why it was more consistently related to lower morbidity risk.