The trace element selenium is of considerable interest due to its toxic and nutritional properties, which markedly differ according to the dose and the chemical form. It has been shown that excess selenium intake increases the risk of type 2 diabetes and, possibly, other metabolic diseases like hyperlipidemia and non-alcoholic fatty liver disease (NAFLD). For the latter, however, epidemiologic evidence is still limited. We carried out a cross-sectional study recruiting 137 healthy blood donors living in Northern Italy and assessed their exposure to selenium through different methods and measuring serum selenium species. We performed linear and spline regression analyses to assess the relation of selenium and its forms with serum alanine aminotransferase (ALT) levels, a marker of NAFLD. Urinary selenium levels were positively and somewhat linearly correlated with ALT (beta regression coefficient (β) 0.11). Conversely, the association of dietary selenium intake with ALT was positive up to 100 µg/day and null above that amount (β 0.03). Total serum selenium was inversely associated with ALT up to 120 µg/L, and slightly positive above that amount. Concerning the different serum selenium species, ALT positively correlated with two organic forms, selenocysteine (β 0.27) and glutathione peroxidase-bound selenium (β 0.09), showed a U-shaped relation with the inorganic tetravalent form, selenite, and an inverse association with human serum albumin-bound selenium (β −0.56). Our results suggest that overall exposure to selenium, and more specifically to some of its chemical forms, is positively associated with ALT, even at levels so far generally considered to be safe. Our findings add to the evidence suggesting that low-dose selenium overexposure is associated with NAFLD.