ObjectiveTo analyze the causal relationship between SARS-CoV-2 infection and retinal vascular obstruction by mendelian randomization (MR). MethodsA two-sample MR analysis utilizing summary statistics from genome-wide association studies (GWAS) in European populations was conducted. The GWAS data for SARS-CoV-2 infection comprised cases of common infection (2 597 856), hospitalized infection (2 095 324), and severe infection (1 086 211). Data on retinal vascular obstruction were obtained from the FinnGen database, which included 203 269 cases of retinal artery obstruction and 182 945 cases of retinal vein obstruction (RVO). Inverse variance weighting (IVW), random effects models, weighted median (WM), MR-Egger regression, simple models, and weighted models were used to analyze the bidirectional causal relationship between different SARS-CoV-2 infection phenotypes and retinal obstruction. The Q statistic was used to assess heterogeneity among single nucleotide polymorphisms (SNP), while MR-Presso was utilized to detect SNP outliers, and MR-Egger intercept tests were performed to evaluate horizontal pleiotropy. ResultsThe MR analysis, using IVW, random effects models, MR-Egger, WM, and weighted models, indicated no significant association between common SARS-CoV-2 infection, hospitalized infection, severe infection, and retinal vascular obstruction (P＞0.05). Additionally, retinal vascular obstruction did not show a significant association with the various SARS-CoV-2 infection phenotypes (P＞0.05). In the simple model, a significant association was found between severe SARS-CoV-2 infection and RVO (P＜0.05), as well as between RVO and common SARS-CoV-2 infection (P＜0.05). No heterogeneity was observed in the IVW and MR-Egger analyses (P＞0.05). The MR-Egger test provided no evidence of horizontal pleiotropy (P＞0.05), and MR-Presso detected no outlier SNP. ConclusionThe findings of this study do not support a causal relationship between SARS-CoV-2 infection and the occurrence of retinal vascular obstruction.