The correlation between subclinical hypothyroidism [thyroid stimulating hormone (TSH)>4 mIU/L with normal free triiodothyroxine and free thyroxine levels], HIV infection and HAART is still unclear.To evaluate the predictive factors of subclinical hypothyroidism in an HIV-infected population, we identified three groups of subjects: G1, subjects on stable highly active antiretroviral therapy (HAART) (for at least 1 year) at baseline and at month 24 (n=97); G2, subjects naive at both baseline and month 24 (n=47); G3, subjects starting HAART at baseline (n=46).The three groups were comparable with respect to age, gender, body weight and prevalence of HCV infection. At baseline, subclinical hypothyroidism was detected in 14 subjects in G1 (14.4%), 5 in G2 (10.6%) and 4 in G3 (8.7%) (P=0.18) and these were excluded from the analysis. At month 24, 15 subjects had developed subclinical hypothyroidism: 4 in G1 (4.8%), 3 in G2 (7.1%) and 8 in G3 (19.0%). In the multivariable analysis, the higher increase in total cholesterol was predictive of subclinical hypothyroidism (RR: 1.53 for each additional 10 mg/dL, 95% CI 1.23-1.90; P<0.01); other variables, which were statistically significant in the univariate analysis, such as G3 group, body weight and higher increase in CD4+ cell count and in triglyceride serum levels were not confirmed to be associated with TSH alterations.The occurrence of subclinical hypothyroidism in HIV-positive patients seems to be related to the increase in total cholesterol serum levels occurring after HAART initiation. Thyroid function should be monitored in all HIV-infected subjects, especially in those starting HAART.