The evidence suggests that trust is an important determinant of health. Trust tends to be lower in low-income and minority individuals, who already suffer from worse health. Therefore, it is particularly important to investigate the predictors of trust in disadvantaged individuals. In this article we use multilevel models to investigate the individual and neighborhood predictors of trust in Mexican-Americans living in low-income neighborhoods (defined as census block groups) in Texas. Detailed survey data on 1754 Mexican-origin respondents provided information on self-rated health and individual characteristics including sociodemographic and sociocultural personal characteristics (frequency of association with people of other races/ethnicities, social support, perceived racism, perceived personal opportunity, and religiosity). Neighborhood heterogeneities and socioeconomic status, computed from census data, were supplemented by community social characteristics (collective efficacy and public disorder) obtained from survey data. Trust was a significant predictor of self-rated health in our sample. This study suggests that Mexican-Americans tend to trust more those with whom there is likely to be a personal acquaintance than other Mexican-Americans. Furthermore, while the results of this study support that people tend to trust more those who are like themselves, for Mexican-Americans, the identification of who is more alike is not based exclusively on racial/ethnic identity, but is a complex process based also on linguistic and socioeconomic similarities. In our sample, linguistic fragmentation, but not racial/ethnic diversity nor neighborhood impoverishment, correlated with trust. Ease of communication seemed to be more important than racial/ethnic homogeneity in encouraging interpersonal trust among Mexican-Americans at the neighborhood level. The findings in this study imply it may be possible to develop neighborhood level interventions, focusing on encouraging social interaction in racially/ethnically and linguistically diverse communities, with the aim of promoting trust to improve health outcomes.