Residential lawns are a dominant and growing feature of US residential landscapes, and the resource-intensive management of this landscape feature presents major potential risks to both humans and the environment. In recent years, scientists and policymakers have been increasingly calling for large-extent measures of lawns and other similar landscape features. Unfortunately, the production of such datasets using traditional, remotely sensed measurement approaches can be prohibitively expensive and time consuming. This study uses two statistical prediction methods to extrapolate the quantity and spatial distribution of residential lawns from a sample of mapped lawns in a large study area in suburban Boston, Massachusetts. The goal is to find an inexpensive, broad-coverage dataset that will provide useable estimates of landscape features in places where we do not have direct measurements of those landscape features. The first estimation method uses OLS regression in conjunction with the sample of mapped lawns and freely available US Census data representing theoretically informed social driver variables. The second, simpler, and less computationally intensive estimation method allocates the mean of the sample of mapped lawns uniformly across the study area. Both estimation methods are performed 1000 times in a Monte Carlo framework where the sample is drawn randomly each realization, to assess the sensitivity of the prediction results to the selection of CBGs in each simple random sample. The outputs of each estimation method are then compared to a reference map where the quantity and spatial allocation of lawns is known for each spatial unit of analysis. Results indicate that the OLS prediction method specified with the independent social driver variables performs better than a uniform prediction method when both are compared to the full-study area reference map.