Gold is possibly the most ancient and, in its recent incarnation as a delivery vehicle for gene therapy, one of the most modern agents in all of medicine's pharmacopoeia. Its administration to humans is both deliberate and inadvertent. It is universally recognized as the most inert of metals, yet it can be sensitizing. Gold's broadest clinical application (in rheumatoid arthritis) derives from a premise that was totally flawed. It is employed clinically to effect immune suppression yet it can engender toxicities that stem from immunostimulation. To complete this series of paradoxes, the toxicity of gold, unlike that of most pharmaceuticals, is, in general, not predictably related to the levels it attains within bodily tissues. Accordingly, the pharmacology and toxicology of gold is remarkably complex. Recent laboratory discoveries concerning gold's metabolism, have emphasized the important metabolic differences between its three oxidation states (0, I and III). When placed in the context of a wealth of clinical experience, these discoveries provide useful insights into its toxicology and shine a revealing light on the mechanisms which account for its seemingly paradoxical behaviour.