Sickle erythrocytes spontaneously generate approximately twice-normal amounts of activated oxygen species (superoxide, peroxide and the highly reactive hydroxyl radical). In addition, they contain excessive amounts of membrane-bound haemichromes, which presumably accumulate because of some degree of sickle haemoglobin instability, perhaps involving interactions between haemoglobin and the lipid bilayer. Since most erythrocyte antioxidant mechanisms are located in the cytoplasm, it is hypothesized that a physiologically significant hydroxyl radical generator would be located within or directly adjacent to the membrane. Indeed, sickle erythrocyte membranes appear to contain a biological 'Fenton reagent', the iron of which is able to cycle between ferrous and ferric states and thereby facilitate superoxide/peroxide-driven hydroxyl radical generation. Data suggest that at least some haemichromes could fulfil this function, although many other potential hydroxyl radical generators exist in sickle erythrocytes. A particularly interesting - but unproven - possibility is membrane-associated haem (without globin). It is further hypothesized that the various membrane abnormalities of sickle erythrocytes might be caused by excessive auto-oxidation. Although proof of this hypothesis does not yet exist, a number of observations are consistent with it. Certainly, evidence indicates that some oxidative modification of sickle erythrocyte membrane proteins and lipids has taken place. Few data exist regarding the state of the erythrocyte membrane in the unstable haemoglobinopathies, but evidence of oxidative perturbation of the HbKöln erythrocyte membrane has been reported. Of great interest is that the membranes of thalassaemic and sickle erythrocytes appear to be remarkably similar in terms of possible oxidative phenomena, suggesting that the membranes of these two cell types may be subjected to similar oxidative stresses.