The U.S. Veterans Health Administration, with oversight of 40 radiation therapy clinics, has begun a program for practice assessment with the goal of identifying, quantifying, and reducing quality variation across its network. Practice assessments will be conducted by teams of surveyors to abstract data from the medical records of patients with a diagnosis of lung or prostate cancer. ASTRO disease-site panels were convened to develop a series of prostate and lung cancer specific quality measures (QMs). For each QM, the set of assessed patients (denominator, exclusions) and the criteria to be satisfied for quality care (numerator) were defined. Additionally, an expected performance rate was specified. An interdisciplinary team of the surveyors, radiation oncologists, medical physicists, and software engineers was convened to review the data sourcing and overall feasibility of the survey process. Software was developed to capture data elements for evaluation of the QMs for each patient cohort. The ASTRO disease site panels generated a series of 27 and 24 QMs for lung cancer and prostate cancer, respectively. The cohorts addressed by the QMs are patients with locally advanced non-small cell lung cancer, limited stage small cell lung cancer, and intermediate and high risk prostate cancer. The QMs span the clinical, dosimetric, and physics components of the medical record. Each QM is designed to be evaluated for each chart under survey for a disease site and to yield pass, fail, or not applicable evaluations. Thus, at an institution level, a passing rate can be calculated for each QM. The disease site panels provided expected institutional performance levels. The survey team evaluated a series of charts at a VA beta site, allowing surveyors to locate, collect, and analyze QMs for prostate and lung cancer cohorts. This process allowed for the design of an efficient software interface and abstraction process. Software to capture metrics for evaluation of QMs has been implemented as a web application with secure transport of anonymized datasets over encrypted connections. The application also supports the remote upload and automatic anonymization of DICOM RT datasets to evaluate dosimetric QMs. Analysis and reporting of results are centralized. After the development of QMs by ASTRO and the implementation of these into a UI, the survey team proceeded to test practice assessments at a beta site. Two surveyors have demonstrated an accrual of 50 cases, at a rate of 20-30 minutes per chart, over the course of a work week. Supporting processes, logistics, and access to target sites have been developed. Initiation of a first site survey is targeted for March of 2017.