Purpose The study aims to support decision making on how best to redesign diabetes care by investigating three potential sources of heterogeneity in effectiveness across trials of diabetes care management. Methods Medline, CINAHL and PsycInfo were searched for systematic reviews and empirical studies focusing on: (1) diabetes mellitus; (2) adult patients; and (3) interventions consisting of at least two components of the chronic care model (CCM). Systematic reviews were analysed descriptively; empirical studies were meta-analysed. Pooled effect measures were estimated using a meta-regression model that incorporated study quality, length of follow-up and number of intervention components as potential predictors of heterogeneity in effects. Results Overall, reviews (n = 15) of diabetes care programmes report modest improvements in glycaemic control. Empirical studies (n = 61) show wide-ranging results on HbA1c, systolic blood pressure and guideline adherence. Differences between studies in methodological quality cannot explain this heterogeneity in effects. Variety in length of follow-up can explain (part of) the variability, yet not across all outcomes. Diversity in the number of included intervention components can explain 8–12% of the heterogeneity in effects on HbA1c and systolic blood pressure. Conclusions The outcomes of chronic care management for diabetes are generally positive, yet differ considerably across trials. The most promising results are attained in studies with limited follow-up (<1 year) and by programmes including more than two CCM components. These factors can, however, explain only part of the heterogeneity in effectiveness between studies. Other potential sources of heterogeneity should be investigated to ensure implementation of evidence-based improvements in diabetes care.