In the field of research on magnetic thin films, interest is rapidly developing in epitaxially grown sandwiches and multilayers because of the multitude of novel magnetic properties which such systems exhibit and because of the possibility they offer for studying certain aspects of fundamental magnetism in a well-controlled manner. One such interesting phenomenon currently under study is the coupling of magnetic layers in metallic superlattices made up of magnetic 3d elements (iron, nickel, cobalt) intercalated with non-magnetic layers. This coupling is found to be antiferromagnetic when the non-magnetic layer is copper or chromium, whereas it would seem to be ferromagnetic in all other cases. Such coupling has been discussed by a number of researchers in terms of Ruderman-Kittel-Kasuya-Yoshida (RKKY) and dipolar interactions but no satisfactory explanation has yet been proposed which gives a full account of the observed behaviour. On simply considering the magnetic response of the non-magnetic layer, ferromagnetic interlayer coupling is predicted for elements such as palladium which are close to the Stoner criterion. Far from the Stoner criterion (copper, gold etc.), a more detailed investigation of the magnetic energy is necessary. In this paper, we propose a novel discussion of the coupling; the magnetic energy due to the hybridization between the magnetic and non-magnetic 0ayers is calculated by a perturbation treatment. Several energy terms are obtained, some of which describe oscillating RKKY-type interactions while others are found to be analogous to antiferromagnetic superexchange interactions in insulators, and may be dominant.