Abstract Background Task-sharing is a promising strategy to expand mental healthcare in low-resource settings, especially in low- and middle-income countries (LMICs). Research on how to best implement task-sharing mental health interventions, however, is hampered by an incomplete understanding of the barriers and facilitators to their implementation. This review aims to systematically identify implementation barriers and facilitators in evidence-based task-sharing mental health interventions using an implementation science lens, organizing factors across a novel, integrated implementation science framework. Methods PubMed, PsychINFO, CINAHL, and Embase were used to identify English-language, peer-reviewed studies using search terms for three categories: “mental health,” “task-sharing,” and “LMIC.” Articles were included if they: focused on mental disorders as the main outcome(s); included a task-sharing intervention using or based on an evidence-based practice; were implemented in an LMIC setting; and included assessment or data-supported analysis of barriers and facilitators. An initial conceptual model and coding framework derived from the Consolidated Framework for Implementation Research and the Theoretical Domains Framework was developed and iteratively refined to create an integrated conceptual framework, the Barriers and Facilitators in Implementation of Task-Sharing Mental Health Interventions (BeFITS-MH), which specifies 37 constructs across eight domains: (I) client characteristics, (II) provider characteristics, (III) family and community factors, (IV) organizational characteristics, (V) societal factors, (VI) mental health system factors, (VII) intervention characteristics, and (VIII) stigma. Results Of the 26,935 articles screened (title and abstract), 192 articles underwent full-text review, yielding 37 articles representing 28 unique intervention studies that met the inclusion criteria. The most prevalent facilitators occur in domains that are more amenable to adaptation (i.e., the intervention and provider characteristics domains), while salient barriers occur in domains that are more challenging to modulate or intervene on—these include constructs in the client characteristics as well as the broader societal and structural levels of influence (i.e., the organizational, mental health system domains). Other notable trends include constructs in the family and community domains occurring as barriers and as facilitators roughly equally, and stigma constructs acting exclusively as barriers. Conclusions Using the BeFITS-MH model we developed based on implementation science frameworks, this systematic review provides a comprehensive identification and organization of barriers and facilitators to evidence-based task-sharing mental health interventions in LMICs. These findings have important implications for ongoing and future implementation of this critically needed intervention strategy, including the promise of leveraging task-sharing intervention characteristics as sites of continued innovation, the importance of but relative lack of engagement with constructs in macro-level domains (e.g., organizational characteristics, stigma), and the need for more delineation of strategies for task-sharing mental health interventions that researchers and implementers can employ to enhance implementation in and across levels. Trial registration PROSPERO CRD42020161357