I n this article the properties of 3d transition metal impurities in simple metal hosts are summarized.T h e microscopic models of Friedel, Anderson and Wolff are reviewed, and the question of the formation of the magnetic moment on the impurity is briefly discussed. The structure of the Anderson model in the Hartree-Fock approximation and the correlation effects in the weakly magnetic limit when the Coulomb correlation energy U is smaller than the width A of the d states are summarized, together with the perturbational treatments of the model in the strongly magnetic limit UiA $1.The s-d exchange model and the Kondo effect (the logarithmic increase of the resistivity with decreasing temperature) are discussed, and the various approximations of this model are compared. T h e relation between the Anderson model and the s-d exchange model is derived, and the nature of the Schrieffer-Wolff transformation is analysed, in particular the question of the polarization of the s and d states.The experimental state of affairs is reviewed next, and the experimental dataobtained mainly on alloys of noble metals and aluminium with 3d elements-are collected to answer the questions raised in the previous sections. Experimental evidence on the singlet ground state, on charge neutrality and on the single-particle and many-body resonances is summarized. The question of the validity of the various approximations of the models is answered by comparing the experimentally found behaviours with theoretically derived formulae; this comparison leads to an indication of the basic unresolved questions in this field.Finally, the latest developments in the theory are reviewed. These developments involve the consideration of the Kondo problem as a typical case of infrared divergencies, and are based on scaling and renormalization group techniques ; they may bridge the gap between the earlier theories and the experimental facts.