We evaluate determinants of community gardens in Philadelphia census tracts by developing a model of community gardening and testing it with negative binomial regression techniques. We find that home vacancy rates, labor force participation rates, poverty rates, and the number of healthy food stores have a positive impact. Theft rates, unemployment rates, the percentage of African Americans and non-citizens, home ownership rates, assault rates, and the existence of parkland all have a negative effect. Population density has an inverse-U effect. Our results are consistent with the argument that community gardening improves food security and social capital in needy neighborhoods.