Serum thyroid-stimulating hormone (TSH) measurement is the diagnostic cornerstone for primary thyroid dysfunction. There is high inter-individual, but limited intra-individual variation in TSH concentrations, largely due to genetic factors. The currently used wide population-based reference intervals may lead to inappropriate management decisions. A polygenic score (PGS) including 59 genetic variants was used to calculate genetically-determined TSH reference ranges in a thyroid disease-free cohort (N=6,834). Its effect on reclassification of diagnoses was investigated when compared to using population-based reference ranges. Next, results were validated in a second independent population-based thyroid disease-free cohort (N=3,800). Potential clinical implications were assessed in a third independent population-based cohort including individuals without thyroid disease (N=26,321) as well as individuals on levothyroxine (LT4) treatment (N=1,132). PGS was a much stronger predictor of individual TSH concentrations than FT4 (total variance in TSH concentrations explained 9.2-11.1% vs. 2.4-2.7%, respectively) or any other non-genetic factor (total variance in TSH concentrations explained 0.2-1.8%). Genetically-determined TSH reference ranges differed significantly between PGS quartiles in all cohorts, while the differences in FT4 concentrations were absent or only minor. Up to 24.7-30.1% of individuals, previously classified as having subclinical hypo- and hyperthyroidism when using population-based TSH reference ranges, were reclassified as euthyroid when genetically-determined TSH reference ranges were applied. Individuals in the higher PGS quartiles had a higher probability of being prescribed LT4 treatment compared to individuals from the lower PGS quartiles (3.3% in Q1 vs. 5.2% in Q4, Pfor trend =1.7x10-8). Individual genetic profiles have potential to personalize TSH reference ranges, with large effects on reclassification of diagnosis and LT4 prescriptions. As the currently used PGS can only predict approximately 10% of inter-individual variation in TSH concentrations, it should be further improved when more genetic variants determining TSH concentrations are identified in future studies.