Objective: Targeting neighborhood characteristics (e.g., pedestrian infrastructure) is a promising mechanism to increase physical activity (PA). However, the association with neighborhood safety and PA is unclear. An ecological approach, which considers multiple levels of influence, may help clarify the association. The current dissertation bridged the PA and injury prevention fields by assessing multilevel moderators of the association between neighborhood safety and PA. Methods: The three samples were overweight/obese adults from San Diego, CA (Study 1, n=298), adolescents from Seattle, WA and Baltimore, MD (Study 2, n=878) and older adults from Seattle, WA (Study 3, n=367). The PA outcomes were self-reported active transport and leisure-time PA, and objective moderate-to-vigorous PA (MVPA). Study 1 assessed perceived neighborhood safety (e.g., safety from crime, pedestrian safety). Studies 2 and 3 assessed objectively-measured pedestrian streetscape safety. Moderators were from individual/demographic, psychosocial, home and neighborhood environment levels. Results: Fifteen of the 17 significant (p<.10) moderators found across studies were from individual/demographic or psychosocial levels. Though some moderators were shared across outcomes within the same study (e.g., sex and BMI in Study 1 or barriers in Study 2), race/ethnicity was the only significant moderator across studies (Studies 1 & 2). White non-Hispanic participants benefitted most in safe vs. unsafe neighborhoods, but there were few benefits among minorities. In Studies 2 & 3, the utility of the new objective streetscape safety measure was demonstrated. It was significantly positively associated with neighborhood-based PA among older adults, but negatively associated among adolescents. Conclusion: The association of neighborhood safety with PA was complex and varied by subgroup. The findings demonstrate the utility of an ecological approach to better understand PA, particularly exploring cross-level interactions. Both perceived and objectively-assessed neighborhood safety had subgroup-specific effects, suggesting each age-group should be targeted separately (e.g., targeting pedestrian safety for older adults). More research is needed that incorporates objectively-assessed pedestrian safety, and interventions should tailor based on individual/demographic and psychosocial characteristics. Interventions can modify safety-related streetscape characteristics to reduce or eliminate traffic-related fatalities and injuries among pedestrians, and may have a dual benefit of facilitating greater physical activity in neighborhoods.