Introduction: The aim of this study was to assess the association between thyroid function and perceived anxiety in non-depressed older adults. Methods: Non-depressed Alzheimer's Disease Neuroimaging Initiative participants with complete Thyroid Stimulating Hormone (TSH) and neuropsychiatric inventory (NPI/NPI-Q) were included. The association between anxiety and thyroid function was assessed by logistic regression and sex stratification. Restricted cubic splines were applied to evaluate non-linearity in the association. Results: The median age of 2,114 eligible participants was 73 years (68-78), 1,117 (52.84%) were males, and the median TSH was 1.69 uIU/mL. There was a significant association between TSH and informant-perceived anxiety in the total study population (OR Model1=0.86, 95% CI 0.76-0.97, p=0.011), even after adjusting for bio-demographical (adj.OR Model2=0.85, 95% CI 0.75-0.96, p=0.007), and socio-cognitive confounders (adj.OR Model3=0.84, 95% CI 0.73-0.96, p=0.009). Sex-stratification showed similar significant results in all models only in males (OR Model1-male=0.71, 95% CI: 0.58-0.85, p Model1-male<0.001). In the general population and males, a TSH value of 2.4 uIU/dL was a significant cutoff under which anxiety odds were significantly high, even after adjusting for confounders. Conclusions: The sex-dependent association between TSH levels and perceived anxiety in non-depressed older adults is a novel finding that has to be further explored for a better understanding of the underlying neurobehavioral biology.