This paper examines the challenges of projecting the long-run sustainable demand for new residential construction and presents a range of estimates for the likely demand for new housing over the period from 2005 through 2014. Making long-run projections requires assumptions about the three elements of the demand for new homes: 1) the demand for additional units to accommodate household growth, 2) the demand for new units to replace existing units lost on the net from the stock, and 3) the demand for additional second homes and vacant units for rent or sale that accommodate the normal turnover of larger housing stock. Assumptions about the future are usually based on examining past trends, as well as judgments about how the future is likely to deviate from these trends as a result of social and economic factors, the age and composition of the housing stock, expected immigration, and expected changes in the size and age distribution of the adult population. Thus, in evaluating any projection it is important that its users understand these assumptions and make their own judgments about the validity of the assumptions and how much faith to put in the projections. Doing so requires not only knowing the assumptions behind the projections but also knowing and understanding the datasets upon which these assumptions and projections are based. Unless the appropriate datasets are selected and handled properly, interpretation of past trends and extrapolations to the future can be faulty.