Background: Genome-wide association studies (GWAS) emerged with expectations borrowed from Mendelian genetics, creating a gap between what the method can and cannot deliver. This gap has fuelled both unwarranted dismissal and unwarranted enthusiasm. For endocrine clinicians, the confusion is consequential because GWAS and clinical sequencing answer fundamentally different questions and serve different purposes.

Purpose: This review defines what GWAS is designed to do, clarifies the state of the "missing heritability" debate in 2026, and evaluates the contribution of GWAS to endocrine and cardiometabolic disease biology, including thyroid function, reproductive timing, adrenal steroid traits, type 2 diabetes, and lipid disorders. It also addresses methodological advances in the modern GWAS toolkit, the current status of polygenic risk scores, and the lessons from applying GWAS to understudied vascular diseases with endocrinological relevance.

Key findings: GWAS maps polygenic architecture, prioritizes biological pathways and tissues, and supports therapeutic target validation using human genetic evidence. It does not provide molecular diagnoses in individual patients. Large meta-analyses have identified hundreds of robust loci for thyroid-stimulating hormone, age at menarche, age at natural menopause, cortisol binding, primary aldosteronism, and type 2 diabetes subtypes. Multi-ancestry designs improve discovery and fine-mapping resolution. In rare but well-phenotyped vascular conditions such as spontaneous coronary artery dissection and fibromuscular dysplasia, GWAS has identified biologically coherent loci and tested hormonal hypotheses, demonstrating that rigorous phenotyping can partly compensate for modest sample sizes. Polygenic risk scores can identify individuals at high genetic risk for common endocrine traits, but portability across ancestry groups and clinical implementation remain unresolved challenges.

Conclusions: GWAS remains a relevant and productive tool in endocrine genetics in 2026, provided its outputs are interpreted correctly. Its role is interpretive and mechanistic rather than diagnostic. Integrated with sequencing, functional genomics, and Mendelian randomization, it offers a structured framework for moving from statistical association to biological understanding and, in selected contexts, to risk stratification research.