Aspects of the method developed by Aaron and McGuire and Maital (AMM) for estimating the benefit distributions associated with the provision of pure public goods are clarified and the method is extended to cover the case of public projects designed to improve the quality of impure public goods. The extended method is used to estimate the distributional impact of a proposed public project in the Cleveland-Akron metropolitan region, and the results are compared with estimates obtained from three naive models and from the AMM model for pure public goods. It is found that the native models overestimate substantially the net benefits received by low-income families and underestimate the net benefits received by high-income families. Using the AMM method for a pure public good similarly distorts the results, though by a much smaller magnitude.