BACKGROUND: Although carbohydrate counting (CC) being the recommended dietary strategy for achieving glycemic control in people with diabetes, there is limited evidence. Our aim is to systematically assess the efficacy of CC in people with diabetes. METHODS: We searched PubMed up to October 2024 and assessed randomized controlled trials of interventions longer than 12 weeks in people with diabetes. Change in glycated hemoglobin (HbA1c) levels was the primary outcome. The results of clinically and statistically homogenous studies were pooled and meta-analyzed using the random-effects model to provide estimates of the efficacy of CC. RESULTS: We identified 16 studies (633 children and 640 adults) in type 1 diabetes mellitus (T1DM) and 6 studies (966 adults) in type 2 diabetes mellitus (T2DM). There was significant improvement in HbA1c levels with CC versus the other diets in T1DM (- 0.21%, 95% CI - 0.41 to - 0.01;  = 0.042) with large heterogeneity (I2 = 85.2%) in people with T1DM. In subgroup analyses, advanced CC (ACC) showed improved HbA1c levels (- 0.47%, 95% CI - 0.78 to - 0.15;  = 0.004) in adults with T1DM, but ACC did not in children with T1DM (- 0.07%, 95% CI - 0.25 to 0.10;  = 0.419). The effects of basic CC (BCC) on glycemic control for children with T1DM were not significant. Additionally, BCC did not show improved HbA1c levels in in adults with T2DM. CONCLUSIONS: For glycemic control in people with T1DM, CC was an effective option. Although BCC was not effective for glycemic control in adults with T2DM, further high-quality and long-term studies are needed to confirm these issues.