Teachers and paraprofessionals need effective training to improve their implementation of interventions for students with disabilities. Reviews of the single-case design literature have identified some features associated with effective training for these educators, but the group-design literature has received little attention. This meta-analysis systematically reviews group-design studies testing the efficacy of training to improve implementation of interventions for students with disabilities. The mean effect size of educator training on implementation fidelity was g = 1.08, and results from meta-regression analysis suggest training that involves a combination of two specific training strategies (i.e., modeling and performance feedback) was associated with improved implementation fidelity. Increased duration of training was not associated with larger effects. Considered alongside findings from the single-case design literature, these results suggest that "how" educators are trained is a more important consideration than the number of hours they spend in training.