This meta-analysis synthesized the findings of experimental and quasi-experimental studies to determine the efficacy of math interventions for enhancing the word problem-solving accuracy of students with mathematics difficulties in Grades 4 to 12. Further, we identified intervention and design characteristics that moderate the effectiveness of interventions. We appraised study quality using Council for Exceptional Children (CEC's) evidence standards. We also used random-effects meta-regression models with robust variance estimation (RVE) to compute the overall weighted treatment effect and investigate the moderating impact of 15 variables. Results from 36 intervention studies published between 1989 and 2020 reveal a relatively large and positive mean effect size (g = 0.71; SE = 0.11, 95% CI [0.49, 0.92]) across 88 outcomes, suggesting the examined interventions were effective. The moderator analyses' findings showed the treatment efficacy varied as a function of five intervention characteristics, including intervention model, interventionist, grade level, math topic, and intervention duration. None of the design features we examined led to significant moderator effects. We discuss findings and limitations and provide directions for future research and practice.