This study aimed to prospectively evaluate whether TSH levels at baseline were associated with incident depression after four years of follow-up in a cohort of middle-aged adults, the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil).TSH and free-thyroxine (FT4) levels were evaluated at baseline. Depression diagnoses were performed using the Clinical Interview Schedule - Revised (CIS-R) at baseline and after a 4-year-follow-up. Poisson regression models (95% Confidence Intervals) were built to evaluate the association between TSH quintiles at baseline and incident depression. All analyses were stratified by sex. Models were presented crude, adjusted for age and sex; and further adjusted for race, education, BMI, smoking, alcohol consumption, use of antidepressants/benzodiazepines, kidney function and comorbidities.Mean age was 51.5 years and 51.2% were women. Overall, low TSH levels (1st quintile) were associated with incident depression (adjusted RR=1.36, 95% CI 1.02-1.81), remaining significant for women (adjusted RR=1.64, 95% CI 1.15-2.33), but not for men. The same results were found when restricting analysis to euthyroid participants (adjusted RR=1.46, 95% CI 1.08-1,99), also significant for women only (adjusted RR=1.63, 95% CI 1.12-2.38).Our results showed that low TSH levels were positively associated with incident depression, particularly among women. Similar results were found when restricting the analysis to euthyroid participants. In contrast, high TSH levels were inversely associated with incident depression, also among women.