Assertive community treatment is known for improving consumer outcomes, but is difficult to implement. On-site fidelity measurement can help ensure model adherence, but is costly in large systems. This study compared reliability and validity of three methods of fidelity assessment (on-site, phone-administered, and expert-scored self-report) using a stratified random sample of 32 mental health intensive case management teams from the Department of Veterans Affairs (VA). Overall, phone, and to a lesser extent, expert-scored self-report fidelity assessments compared favorably to onsite methods in inter-rater reliability and concurrent validity. If used appropriately, these alternative protocols hold promise in monitoring large-scale program fidelity with limited resources.