Our paper provides an empirical assessment of the effectiveness of pre-purchase homeownership counseling in reducing 90-day delinquency rates. We use data on nearly 38,000 fixed-rate, purchase money mortgages originated under Freddie Mac’s affordable lending programs between the years 2000 and 2008. We take efforts to control for the quasi experimental nature of our data, as well as the heterogeneous experience of borrowers post origination. We find that counseling reduces the delinquency rate of first-time home buyers by 29 percent, that counseling’s effectiveness is largely insensitive to its method of delivery, and that its effectiveness was greatest in the boom/crisis years of 2005 through 2008. We estimate the dollar benefit of counseling’s reduction in delinquency rates to be about $1,000, easily sufficient to pay for its delivery. The recent housing crisis has raised alarm over high foreclosure rates and the sustainability of homeownership. This concern is particularly relevant for low-income and first-time home buyers, who were especially hard hit in the crisis. Housing advocates have called for the development and establishment of policies that will mitigate default risk while allowing the extension of credit to these targeted populations. It is within this content that we direct our attention to the efficacy of homeownership counseling. Pre-purchase homeownership counseling has frequently been integrated with efforts to responsibly extend credit to borrowers. Counseling’s proponents have argued that it reduces the probability of borrowers becoming delinquent or going into default. However support for counseling has been tempered by the inconclusive empirical evidence of its efficacy. Our hope is to provide more decisive evidence in this paper. The absence of definitive evidence of the benefits of pre-purchase homeownership counseling is not from a lack of trying, there is rich literature on the subject. Existing literature is generally found deficient on two dimensions; relying on a small sample, and/or relying on data derived from a quasi-experiment and failing to appropriately account for the associated potentials for bias and inconsistency.