Objective: To examine variation by race and gender in the association between neighborhood socioeconomic status and walking among community-dwelling older adults. Design: Cross-sectional. Setting: Cardiovascular Health Study, a longitudinal population-based cohort. Participants: 4,849 adults, aged > 65 years Measurements: Participants reported the number of city blocks walked in the prior week. Neighborhood socioeconomic status (NSES) was measured at the level of the census tract. Negative binominal regression models were constructed to test the association between NSES and blocks walked. In the fully adjusted models, we included two-way and three-way interaction terms among race, gender, and NSES. Results: In adjusted analyses, among White residents in the lowest NSES quartile (most disadvantaged), men walked 64% more than women (P<.001), while in the highest NSES (most advantaged), men walked 43% more than women (P<.001). Among African American residents in the lowest NSES quartile, men walked 196% more blocks than women (P<.001). Conclusions: Female gender is more strongly associated with walking for African Americans than for Whites in low SES neighborhoods but had a similar association with walking for both African Americans and Whites in high SES neighborhoods.