Evidence-based programs (EBPs) delivered in elementary schools show great promise in reducing risk for emotional and behavioral disorders (EBDs). However, efforts to sustain EBPs in school face barriers. Improving EBP sustainment thus represents a priority, but little research exists to inform the development of sustainment strategies. To address this gap, the Sustaining Evidenced-Based Innovations through Multi-level Implementation Constructs (SEISMIC) project will (a) determine if malleable individual, intervention, and organizational factors predict EBP treatment fidelity and modifications during implementation, sustainment, or both; (b) assess the impact of EBP fidelity and modifications on child outcomes during implementation and sustainment; and (c) explore the mechanisms through which individual, intervention, and organizational factors influence sustainment outcomes. This protocol article describes SEISMIC, which builds upon a federally funded randomized clinical trial evaluating BEST in CLASS, a teacher-delivered program for K to third grade children at risk for EBDs. The sample will include 96 teachers, 384 children, and 12 elementary schools. A multi-level, interrupted time series design will be used to examine the relationship between baseline factors, treatment fidelity, modifications, and child outcomes, followed by a mixed-method approach to elucidate the mechanisms that influence sustainment outcomes. Findings will be used to create a strategy to improve EBP sustainment in schools.