BackgroundNon-proliferative diabetic retinopathy (NPDR), a common diabetic complication with high morbidity, is featured by impaired visual function and fundus lesions. It has been reported that oral Chinese patent medicines (OCPMs) may improve visual acuity and fund signs. However, the best possible OCPMs for NPDR remain questionable and merit further investigation.MethodsFrom inception to October 20, 2022, seven databases were searched for eligible randomized controlled trials (RCTs). The outcomes were clinical effective rate, visual acuity, visual field gray value, microaneurysm volume, hemorrhage area, macular thickness, and adverse events rate. The revised Cochrane risk-of-bias tool (ROB 2) was used to assess the quality of the included studies. Network meta-analysis was performed using R 4.1.3 and STATA 15.0 software.ResultsWe included 42 RCTs with 4,858 patients (5,978 eyes). The Compound Danshen Dripping Pill (CDDP) combined with calcium dobesilate (CD) had the most improvement in clinical efficacy rate (SUCRA, 88.58%). The Compound Xueshuantong Capsule (CXC) combined with CD may be the best intervention (SUCRA, 98.51%) for the improvement of visual acuity. CDDP alone may be the most effective treatment option (SUCRA, 91.83%) for improving visual field gray value. The Hexuemingmu Tablet (HXMMT) and Shuangdan Mingmu Capsule (SDMMC) combined with CD may be the most effective treatment for reducing microaneurysm volume and hemorrhage area (SUCRA, 94.48%, and 86.24%), respectively. Referring to reducing macular thickness, CXC combined with CD ranked first (SUCRA, 86.23%). Moreover, all OCPMs did not cause serious adverse reactions.ConclusionOCPMs are effective and safe for NPDR. CDDP alone, and combined with CD, may be the most effective in improving visual field gray value and clinical efficacy rate, respectively; CXC combined with CD may be the best in enhancing BCVA and reducing macular thickness; HXMMT and SDMMC combined with CD, maybe the most effective regarding microaneurysm volume and hemorrhage area, respectively. However, the reporting of methodology in the primary study is poor, potential biases may exist when synthesizing evidence and interpreting the results. The current findings need to be confirmed by more large-sample, double-blind, multi-center RCTs of rigorous design and robust methods in the future.Systematic review registrationhttps://www.crd.york.ac.uk/prospero/, identifier CRD42022367867.