Blended learning integrating asynchronous online modules with synchronous peer interaction shows promise for science teacher education, yet optimal synchronization patterns remain understudied. This systematic review synthesized empirical evidence on blended learning designs in science teacher education, examining synchronization models, learning outcomes, and implementation factors. Following PRISMA 2020 guidelines, five databases (ERIC, Scopus, Web of Science, PsycINFO, and Education Source) were searched, yielding 38 empirical studies published between 2013 and 2024. Quality appraisal used ROBINS-I and CASP tools; findings were organized through narrative synthesis by synchronization model, discipline, and geographic context. Three synchronization patterns emerged: sequential (42%), parallel (37%), and integrated (21%). Sequential models, where asynchronous foundational content preceded synchronous peer microteaching, demonstrated the strongest gains in reflective competency, collaborative skills, and digital literacy. Critical success factors included structured peer-feedback rubrics, facilitator modeling during synchronous sessions, and explicit learning pathways that connect both modalities. Programs in resource-constrained contexts achieved comparable outcomes when offline-accessible asynchronous content and strategically scheduled synchronous sessions were implemented. Physics education particularly benefited from simulation-based asynchronous instruction paired with synchronous pedagogical discussion. This review concludes that intentional synchronization, supported by sound theoretical design, structured feedback systems, and attention to infrastructure constraints, significantly enhances science teachers' reflective competency, collaborative skills, and digital literacy. Future research should prioritize longitudinal studies tracking competency transfer to classroom practice, head-to-head model comparisons, and cost-effectiveness analyses.