The pathogenesis of ocular diseases (ODs) remains unclear, although genome-wide association studies (GWAS) have identified numerous associated genetic risk loci. We integrated protein quantitative trait loci (pQTL) datasets and five large-scale GWAS summary statistics of ODs under a cutting-edge systematic analytic framework. Proteome-wide association studies (PWAS) identified plasma and brain proteins associated with ODs, and 11 plasma proteins were identified by Mendelian randomization (MR) and colocalization (COLOC) analyses as being potentially causally associated with ODs. Five of these proteins (protein-coding genes ECI1, LCT, and NPTXR for glaucoma, WARS1 for age-related macular degeneration (AMD), and SIGLEC14 for diabetic retinopathy (DR)) are newly reported. Twenty brain-protein–OD pairs were identified by COLOC analysis. Eight pairs (protein-coding genes TOM1L2, MXRA7, RHPN2, and HINT1 for senile cataract, WARS1 and TDRD7 for AMD, STAT6 for myopia, and TPPP3 for DR) are newly reported in this study. Phenotype-disease mapping analysis revealed 10 genes related to the eye/vision phenotype or ODs. Combined with a drug exploration analysis, we found that the drugs related to C3 and TXN have been used for the treatment of ODs, and another eight genes (GSTM3 for senile cataract, IGFBP7 and CFHR1 for AMD, PTPMT1 for glaucoma, EFEMP1 and ACP1 for myopia, SIRPG and CTSH for DR) are promising targets for pharmacological interventions. Our study highlights the role played by proteins in ODs, in which brain proteins were taken into account due to the deepening of eye–brain connection studies. The potential pathogenic proteins finally identified provide a more reliable reference range for subsequent medical studies.