Experimental studies have consistently shown that iron is a critical catalyst in generating oxygen free radicals via Fenton chemistry. Nevertheless, epidemiologic studies conflict on the association between stored body iron markers and disease outcomes, including coronary heart disease. We hypothesize that stored body iron markers common in epidemiologic studies, such as serum ferritin, transferrin saturation, iron, or iron-binding capacity, are inappropriate to investigate harmful health effects related to iron overload. Oxygen free radicals are produced only by free iron, but stored body iron markers reflect iron bound to ferritin or transferrin, which are produced to sequester catalytically active free iron. Moreover, increased serum ferritin may occur as a defense mechanism in response to oxidative stress; such increase might eventually minimize oxidative stress and consequent pathology due to free iron. Therefore, though highly correlated with stored body iron, a measure of bound iron will fail to identify any harmful effect, unless it is also a marker of free iron. It is generally believed that free iron rarely exists, except in iron-overload with 100% transferrin saturation. However, some recent studies find non-transferrin bound iron (NTBI) or the intracellular labile iron pool (LIP) in the presence of triggers disturbing iron homeostasis, such as alcohol consumption. In contrast to the tight bond in ferritin or transferrin, free iron is more likely to dissociate from a looser bond. Therefore research on the relation of iron with disease outcomes should investigate NTBI or the intracellular LIP. Any positive influence of iron on coronary heart and other diseases might be observable only when a trigger is present. These factors may explain why there have been conflicting results between serum markers of stored body iron and disease outcomes in epidemiological studies.