A transgenic mouse model expressing exclusively haemoglobin S has been an active objective of sickle cell anaemia research in the last decade because it can: (i) help to understand the pathophysiology of the disease; (ii) help in the development of anti-sickling drugs; and (iii) have potential usefulness for dry runs of anti-sickling gene therapy.Three groups have reported sickle transgenic mouse lines that express only human globin chains Ryan et al, 1997;Chang et al, 1998). This accomplishment required the`knockout' of the murine globin genes, that is, the removal or inactivation of these genes by homologous recombination (Ciavetta et al, 1995;Pa Ászty et al, 1995;Yang et al, 1995;Chang et al, 1998). The strategy used to generate the`knockout' models is described in Fig 1. A better perspective on the characteristics of knockout models with exclusively human globin chains is obtained by first discussing the sickle transgenic models previously available (which are still useful for defined purposes) and contrasting them with the new models.Models with a combination of murine globins and human globin chains The first models expressing moderately high levels of both human a-and b S -globin chains were reported by Greaves et al (1990) and Ryan et al (1990). The former was not propagated and the latter expressed only about 50% b S . Subsequently, many different lines of mice with variable degrees of severity became available allowing more detailed studies of pathology.The SAD-1-mouse. This mouse, generated/characterized by Trudel et al (19911994), expresses the SAD transgene at 19% that contains three mutations (b S , b S±Antilles and b DLos Angeles or b SAD ) in the same b chain (Trudel et al, 1991). The last two mutations enhance polymer formation. When the b major deletion (Skow et al, 1983) is present as the hemizygote (b-thal/SAD or SAD-1/b d3 ), the percentage of b SAD increases to 26% and the phenotype is severe. A human a-globin gene is co-integrated at the same site. This model has been used to test clotrimazol, a Gardos channel blocker that inhibits dense-cell formation and to test Mg 11 supplementation (De Franceschi et al, 1994. The model has provided information on glomerular abnormalities and could be useful for testing therapeutic approaches for priapism. Recently, ineffective erythropoiesis has been demonstrated in SAD and SAD-1/ b d3 mice (Blouin et al, 1999).Blouin et al (2000) have recently described the effect of introducing several levels of human g on the SAD-1 mouse that has no deletions or knockouts. The effect of g on transgenic mouse models is important for both understanding the mechanism of the protective effect of g and as a baseline for evaluating anti-sickling haemoglobins proposed for use in gene therapy. Four different levels of human g were bred into the SAD-1 mouse resulting in mice expressing 5´2, 10´6, 18´4 and 24´4% g as a percentage of all b-like chains; there was a concomitant decrease in the percentage of SAD haemoglobin that was not quantified. The effect of the SAD transgene...