Housing First (HF) combines permanent supportive housing and supportive services for homeless individuals and removes traditional treatment-related preconditions for housing entry. There has been little research describing strengths and shortfalls of HF implementation outside of research demonstration projects. The U.S. Department of Veterans Affairs (VA) has transitioned to an HF approach in a supportive housing program serving over 85,000 persons. This offers a naturalistic window to study fidelity when HF is adopted on a large scale. We operationalized HF into 20 criteria grouped into 5 domains. We assessed 8 VA medical centers twice (1 year apart), scoring each criterion using a scale ranging from 1 (low fidelity) to 4 (high fidelity). There were 2 HF domains (no preconditions and rapidly offering permanent housing) for which high fidelity was readily attained. There was uneven progress in prioritizing the most vulnerable clients for housing support. Two HF domains (sufficient supportive services and a modern recovery philosophy) had considerably lower fidelity. Interviews suggested that operational issues such as shortfalls in staffing and training likely hindered performance in these 2 domains. In this ambitious national HF program, the largest to date, we found substantial fidelity in focusing on permanent housing and removal of preconditions to housing entry. Areas of concern included the adequacy of supportive services and adequacy in deployment of a modern recovery philosophy. Under real-world conditions, large-scale implementation of HF is likely to require significant additional investment in client service supports to assure that results are concordant with those found in research studies.