Background: Linking large-scale population-based biobanks with electronic health records allows researchers to study the genetics of complex diseases on an unprecedented scale. Nevertheless, large-scale ICD-10-based analyses remain uncommon, leaving many population-specific genetic signals and shared disease mechanisms undiscovered. We conducted genome-wide association analyses of 5,522 ICD-10–defined phenotypes (4,884 non-sex-specific, 546 female-specific, and 62 male-specific) in 206,159 Estonian Biobank participants using REGENIE, analysing 18.9 million imputed variants. We identified lead variants for each genome-wide significant locus, and we assessed novelty through overlap with the GWAS Catalog and Open Targets, and fine-mapped putatively novel leads to refine likely causal signals. Post-GWAS analyses included SNP heritability estimation and cross-trait genetic and phenotypic correlation analyses. Phenotypes from the ICD-10 category “Endocrine, nutritional, and metabolic diseases” had the highest number of significant loci, followed by “Skin and subcutaneous tissue” and “Circulatory system”. We identified 3372 genome-wide significant loci linked to 1033 phenotypes, including 176 putatively novel associations. Significant heritability was observed for 297 phenotypes, with strong genetic and phenotypic correlations among cardiovascular, endocrine, and immune-related domains. Interestingly, we also observed a robust genetic correlation with procedural “Non-diagnostic medical codes”, emphasising the clinical trajectory from disease to intervention. We furthermore performed in-depth analysis of the causal single-nucleotide variations. Despite the limitations of ICD-10-based phenotyping, this study demonstrates the strength of national biobanks for robust, large-scale disease mapping. Our findings expand the catalogue of disease-associated loci in the Estonian population and underscore the value of deeply phenotyped, population-specific cohorts for uncovering genetic risk factors in human disease.