Background: Subclinical hypothyroidism has been associated with depressive symptoms in cross-sectional studies, but prospective data and data on subclinical hyperthyroidism are scarce. Methods: In the Leiden substudy of the Prospective Study of Pravastatin in the Elderly at Risk (PROSPER), thyroid-stimulating hormone and free T4 levels were measured at baseline and repeated after 6 months in adults aged 70-82 years with preexisting cardiovascular disease or known cardiovascular risk factors to define persistent thyroid functional status. Main outcome measures were depressive symptoms, assessed with the Geriatric Depression Scale 15 (GDS-15) at baseline and after 3 years. All analyses were adjusted for age, gender and education. Results: In 606 participants (41% women; mean age 75 years) without antidepressant medication, GDS-15 scores at baseline did not differ for participants with subclinical hypothyroidism (n = 47; GDS-15 score 1.75, 95% CI 1.29-2.20, p = 0.53) or subclinical hyperthyroidism (n = 13; GDS-15 score 1.64, 95% CI 0.78-2.51, p = 0.96) compared to euthyroid participants (n = 546; mean GDS-15 score 1.60, 95% CI 1.46-1.73). After 3 years, compared to the euthyroid participants, changes in GDS-15 scores did not differ for participants with subclinical hypothyroidism (ΔGDS-15 score -0.03, 95% CI -0.50 to 0.44, p = 0.83), while subclinical hyperthyroidism was associated with an increase in GDS scores (ΔGDS-15 score 1.13, 95% CI 0.32-1.93, p = 0.04). All results were similar for persistent subclinical thyroid dysfunction. Conclusions: In this largest prospective study on the association of persistent subclinical thyroid dysfunction and depression, subclinical hypothyroidism was not associated with increased depressive symptoms among older adults at high cardiovascular risk. Persistent subclinical hyperthyroidism might be associated with increased depressive symptoms, which requires confirmation in a larger prospective study.