The impact of actinides on living organisms has been the subject of numerous studies since the 1950s. From a general point of view, these studies show that actinides are chemical poisons as well as radiological hazards. Actinides in plasma are assumed to be mainly complexed to transferrin, the iron carrier protein. This paper casts light on the uptake of actinides(IV) (thorium, neptunium, plutonium) by transferrin, focusing on the pH dependence of the interaction and on a molecular description of the cation binding site in the protein. Their behavior is compared with that of iron(III), the endogenous transferrin cation, from a structural point of view. Complementary spectroscopic techniques (UV/Vis spectrophotometry, microfiltration coupled with gamma spectrometry, and X-ray absorption fine structure) have been combined in order to propose a structural model for the actinide-binding site in transferrin. Comparison of our results with data available on holotransferrin suggests some similarities between the behavior of Fe(III) and Np(IV)/Pu(IV)/ Np(IV) is not complexed at pH <7, whereas at pH approximately 7.4 complexation can be regarded as quantitative. This pH effect is consistent with the in vivo transferrin "cycle". Pu(IV) also appears to be quantitatively bound by apotransferrin at around pH approximately 7.5, whereas Th(IV) was never complexed under our experimental conditions. EXAFS data at the actinide edge have allowed a structural model of the actinide binding site to be elaborated: at least one tyrosine residue could participate in the actinide coordination sphere (two for iron), forming a mixed hydroxo-transferrin complex in which actinides are bound with transferrin both through An-tyrosine and through An--OH bonds. A description of interatomic distances is provided.