Urban parks are usually studied as discrete, green public open spaces. Less studied is how parks are geographically distributed from a 'spatial logic' point of view, i.e. how they ought to be geographically distributed across the urban landscape. This paper evaluates the degree to which normative principles about park distribution are in evidence from the standpoint of three spatial goals: proximity, diversity and social need. Using Phoenix and Chicago as case studies, it offers an empirical example of how park distribution can be shown to conform (or not) to one or more distributive patterns. Descriptive measures show that Phoenix parks do not conform to any of these three distributional principles, while Chicago parks do somewhat better. In both cities, land uses surrounding parks heavily favour residential uses. In Phoenix, single-family homes are prioritized, making spatial distribution based on proximity, diversity or social need difficult. A strategy for acquiring a better spatial logic, within existing realities and constraints, is suggested.