Individuals decide to use public transit in part on the basis of their perception of transit safety, which is determined by various individual and environmental factors. This paper adopts a multilevel approach to analyze how perceptions of bus and train safety in Chicago, Illinois, vary as a function of person-level characteristics - gender, age, ethnicity, income, and frequency of transit ridership - and neighborhood-level characteristics - perceived neighborhood disorder, population density, and level of poverty. Hierarchical linear modeling is applied to a unique data set that combines data from three different sources: Chicago Transit Authority data on individual-level perceptions of transit safety and individual demographics, Chicago Alternative Policing Strategy data on perceptions of neighborhood disorder at the zip code level, and U.S. Census zip code-level demographic data. Findings show that the individualand zip code-level effects differ depending on whether the estimation predicts bus or train safety perceptions. Additionally, while higher-income individuals and African Americans report lower levels of perceived bus safety, there is an additional zip code effect: low-density and high-poverty areas and neighborhoods with high levels of disorder have significantly lower levels of perceived bus safety. Management and policy implications of the study for both the transit agency and the community in which service is being offered are discussed.