Organizations that disperse their research and development (R&D) activities across geographic and functional boundaries must integrate specialized, distributed knowledge. However, the extant research on this challenge remains conceptually fragmented. To develop an integrated, multilayered model of knowledge integration in decentralized R&D management, we synthesized this literature. Using a meta-synthesis method following Sandelowski and Barroso’s seven-step approach, we examined 4437 studies published between 2000 and 2025, narrowed the pool to 241, and purposively selected 70 on the basis of predefined inclusion and exclusion criteria. Through content analysis and synthesis employing open, axial, and selective coding, we identified a total of 331 open codes and 34 axial codes for systematic analysis.
The analysis conceptualizes knowledge integration as a dynamic capability activated by four clusters of interdependent mechanisms: cognitive transformation (translation, interpretation), relational coordination (trust, social exchange), structural networking (boundary spanning, digital platforms), and system-level leadership (governance design, contextual orchestration). Aligning these mechanisms with five contextual domains—innovation space, geographic dispersion, organizational complexity, interorganizational network configuration, and the tacitness of knowledge—mitigates barriers arising from hierarchy, dispersion, and cognitive heterogeneity. The model shows that multinational and cross-functional R&D units can achieve radical innovation by purposively recombining dispersed expertise, and it clarifies pathways for theory development at the individual, team, organizational, and ecosystem levels.