Neighborhood influences on children and youth are the subjects of increasing numbers of studies, but there is concern that these investigations may be biased, because they typically rely on census-based units as proxies for neighborhoods. This pilot study tested several methods of defining neighborhood units based on maps drawn by residents, and compared the results with census definitions of neighborhoods. When residents' maps were used to create neighborhood boundary definitions, the resulting units covered different space and produced different social indicator values than did census-defined units. Residents' agreement about their neighborhoods' boundaries differed among the neighborhoods studied. This pilot study suggests that discrepancies between researcher and resident-defined neighborhoods are a possible source of bias in studies of neighborhood effects.