The spatial arrangement of public open spaces in communities has an important influence on the recreational net benefits from those public open spaces. A prime example of a public open space in communities where spatial arrangement is important is parks. From the perspective of maximizing the net benefits of recreation, there is a tradeoff between placing all the land for parks in a single park and making several parks to reduce the travel costs of households to the parks. If several parks are made, then the amount of land in each park is reduced, and the recreational net benefit of a trip to any of the parks is less. Since recreation is an important source of value from parks, an examination of an optimal spatial arrangement of parks for recreation in a community is of interest to community planners. The community is assumed to be a slice of a larger urban area or a small town since the housing structure and socioeconomic characteristics of the community is assumed homogeneous. The demand for recreation trips to parks is shifted by the socioeconomic characteristics of the population and the size of the parks. The size of the parks is the division of the number of parks into the total amount of land in parks. The price of a trip to a park is the cost of a round trip to the park. The consumer surplus of a trip to a park is the net benefit the person receives from the park. The amount of land for parks and number of parks to maximize the net benefits from recreation is determined from the model of the demand for trips to a park. Comparative static results suggest that the optimal amount of land, number of parks, and the size of the parks depend on the socioeconomic characteristics of the city. Cities with higher populations, more income, and more education should have more land in parks, more parks, and the parks should be smaller. Lower travel costs and prices of land should result in more land in parks and more parks, but there should be no influence on the park size. Data are collected from seventy cities on the amount of land, size, and number of parks. The distances between the parks, distances of the parks from downtown, and other variables relevant to the spatial distribution of parks are also obtained. Data on the socioeconomic characteristics of the population, travel costs, and the price of land are collected for the cities. The same data is collected for each of the one hundred and sixty nine zip codes associated with the seventy cities. These zip codes are spatially smaller and may better represent homogeneous communities. Three equations are estimated to learn the influence of the socioeconomic and travel cost features of a city on the spatial distribution of parks. The three equations are the estimation of the amount of land, the number and the size of parks. The results of the estimation for the amount of land in parks equation match the theoretical predictions, and the fit of the relationship is good. Population, population density, and land price have a significant influence on the amount of land in parks. A larger population makes a city create more land in parks while population density and the price of land makes a city reduce the amount of land. The results of the estimation for the number and size of parks are less clear. The results match the comparative static predictions loosely, and the fit of the equations is not good. These findings may reflect incorrect assumptions on the preferences of people for park size in the theoretical model, or the data that the preferences are supposed to represent are clouded with noise from government regulations, costly removal of parks, and other features of the complexity of urban spatial structure. Although recreation is significant component of the value households derive from parks, there has been little attention paid to determining the spatial arrangement of parks to maximize the net benefits. The theoretical results suggest that the amount of land and the spatial distribution of parks should be sensitive to the socioeconomic characteristics of a community. However, only the predicted signs of the socioeconomic characteristic on the amount of land in parks hold up empirically. The spatial distribution of land is more sensitive to institutional factors that probably make it necessary to have more data on parks to fully identify the relationship between the socioeconomic characteristics and the number and size of parks.