Context: In cross-sectional studies, serum TSH concentrations increase with age. This has not been examined longitudinally, and it is uncertain whether the TSH increase reflects healthy aging or occult thyroid failure. Methods: We measured serum TSH, free T4, thyroid peroxidase, and thyroglobulin antibodies in 1100 participants in the 1981 and 1994 Busselton Health Surveys and derived a reference group of 908 individuals without thyroid disease or thyroid antibodies. We examined changes in thyroid function longitudinally and, in 781 participants, explored associations with the CAPZB polymorphism rs10917469. Results: At 13 yr follow-up, mean serum TSH increased from 1.49 to 1.81 mU/liter, a change in mean TSH (ΔTSH) of 0.32 mU/liter 95% confidence interval (CI) 0.27, 0.38, P < 0.001, whereas mean free T4 concentration was unchanged (16.6 vs. 16.6 pmol/liter, P = 0.7). The TSH increase was most marked in the elderly, such that gender-adjusted ΔTSH increased by 0.08 mU/liter (95% CI 0.04, 0.11) for each decade of baseline age. People with higher baseline TSH values had proportionally smaller increases in TSH, with each additional 1.0 mU/liter of baseline TSH associated with a 0.13 mU/liter decrease (age and gender adjusted) in ΔTSH (95% CI 0.09, 0.16). The ΔTSH did not differ significantly by CAPZB genotype. Conclusions: Aging is associated with increased serum TSH concentrations, with no change in free T4 concentrations. The largest TSH increase is in people with the lowest TSH at baseline. This suggests that the TSH increase arises from age-related alteration in the TSH set point or reduced TSH bioactivity rather than occult thyroid disease.