Severe sepsis is characterized by rapid development of multiple organ failure associated with high mortality. Bacterial toxin release triggers a sequence of events that activates intracellular pathways to produce inflammatory mediators and nitric oxide. There have been numerous attempts to interrupt this devastating cascade by removing toxins, removing or inhibiting mediators, and by blocking receptors of mediators. This review considers toxin properties with a strong focus on toxic shock syndrome toxin 1 and the potential of various removal technologies in relation to these properties. The distribution of toxins in vivo forms a key issue but is nevertheless poorly defined. For toxic shock syndrome toxin 1, either a high clearance or a high degree of compartmentalization to a space not accessible by pheresis or immunoabsorption technologies seems likely. Attempts to remove toxins to treat sepsis may appear futile if we cannot access this space or when the level of induced clearance is too low compared with natural clearance. The impact of these considerations is highly dependent on the exact toxin biology in vivo. Extrapolated to other toxins, we indicate a set of general requirements to be met to facilitate successful toxin removal by a pheresis technique.