Purpose: To compare safety and therapeutic effect of three treatment protocols on patients with naive proliferative diabetic retinopathy.Methods: A total of 207 eyes with proliferative diabetic retinopathy were randomly divided into three groups: full panretinal photocoagulation group; intravitreal bevacizumab (IVB) group with four monthly IVB injections; and modified combination group with two bimonthly IVB injections and a modified laser therapy. The best-corrected visual acuity and area of neovascularization leakage were compared at 1-year follow-up.Results: The difference in final best-corrected visual acuity was not significant between the groups (P = 0.77). The modified combination group had the lowest final leakage area (P = 0.006). The difference in final mean deviation of visual field was not significant between IVB and modified combination groups (mean difference = 0.25, P = 0.23, 95% confidence interval, 0.12-1.38). There was no difference in rate of new-onset diabetic macular edema between IVB and modified combination groups (mean difference = 1.5%, P = 0.31, 95% confidence interval, 1.1-1.88). Mean of total IVB injections were 3.5, 7.4, and 6.2 for panretinal photocoagulation, IVB, and modified combination groups, respectively (P = 0.002). Patients in the IVB group underwent more visits (P = 0.001). In subgroup analysis, the difference in the final leakage area was significant for the eyes with diabetic macular edema (P = 0.005).Conclusion: A combination protocol of photocoagulation and IVB can be recommended for proliferative diabetic retinopathy, especially with baseline diabetic macular edema.