Background <p>Previous studies have shown that cognitive behavioral therapy (CBT) is beneficial for enhancing blood sugar monitoring and mental health in diabetic patients, but no studies have compared all CBT formats to determine which one is more effective. This review compared different delivery formats for CBT for diabetes.</p> Methods <p>We searched nine databases for randomized clinical trials comparing one CBT delivery format against another format or control conditions for patients with diabetes. The primary outcomes were glycated hemoglobin (HbA1c) reduction, severity of depression and anxiety. Pairwise and network meta-analyses were used to synthesize the data.</p> Results <p>29 trials with 5,208 participants were included. Network meta-analyses showed at post-intervention, with very low to moderate evidence, digital assisted was probably the most effective in decreasing depression compared with unguided and guided self-help, individual, group, treatment as usual (TAU), education, waitlist plus TAU and waitlist. At follow-up, moderate evidence indicated digital assisted significantly reduced depression and anxiety compared to group, education, TAU, and waitlist, and outperformed individual and waitlist plus TAU for depression. Evidence from very low to low indicated other CBT formats also showed improvement in relevant outcomes. MD range for glycated hemoglobin (follow-up): -1.24 to -0.45%; MD range for depression (post-intervention): -35.08 to -8.21; MD range for depression (follow-up): -54.00 to 5.89; SMD range for anxiety: -17.51 to -0.74; SMD range for quality of life: 2.12 to 3.39.</p> Discussion <p>Digital-assisted CBT appears most effective for reducing depression and anxiety, while group and guided self-help formats showed benefits for HbA1c and distress. Limitations include potential publication bias on group CBT format for HbA1c and distress, lack of anxiety outcomes and digital assisted format studies.</p> Registration <p>PROSPERO (CRD42024515804).</p>