It is difficult to get evidence from randomized trials of a causal relationship between steroid hormones produced by the adrenal gland and gonad and retinal neurodegenerative disorders (RND). In this study, genetic variations of aldosterone (Aldo), androstenedione (A4), progesterone (P4), hydroxyprogesterone (17-OHP), and testosterone/17β-estradiol (T/E2) were obtained from genome-wide association studies as instrumental variables. Mendelian randomization (MR) analysis was used to assess the impact on the risk of RND, including glaucoma (8,591 cases and 210,201 controls), diabetic retinopathy (DR, 14,584 cases and 202,082 controls) and age-related macular degeneration (AMD, 14,034 cases and 91,214 controls). As the main method, inverse variance weighted results suggest that the increased glaucoma risk was affected by T/E2 (OR = 1.11, 95% CI, 1.01–1.22, P = 0.03), which was further validated by other methods (PWM= 0.03, PMLE= 0.03, PMR-RAPS= 0.03). In the replicated stage, the causal relationship between T/E2 and glaucoma was verified based on the MRC-IEU consortium (P = 0.04). No impact of Aldo, A4, P4, 17-OHP, and T/E2 was observed for the risk of DR (P > 0.05) and AMD (P > 0.05). The heterogeneity test (P > 0.05) and pleiotropy test (P > 0.05) verified the robustness of the results. Our results suggest that T/E2 has a suggestive effect on the glaucoma risk. However, the genetic evidence based on a large sample does not support the effect of steroid hormones on DR and AMD risk. Further studies are vital to assess the possibility of steroid hormones as targets for prevention and treatment.