We propose a model of the interaction between the metallic centers of a polymetallic system which permits not only the rationalization of most of the results already known, but also the design of new systems exhibiting expected properties. If we consider a copper(II) dimer, the interaction leads to two low lying states, a spin singlet and a spin triplet, separated by J . The interaction is said to be antiferromagnetic if the spin singlet is lower in energy ( J J > 0). In our model, the energy gap J is the algebraic sum of two components, J AF and J F , favoring the antiferro- and the ferromagnetic situations respectively. In some way, the sign and the magnitude of J result from the opposition between two fighters. These fighters, as the gladiators, have not the same tools. The tools of the antiferromagnetic gladiator are the overlap S between the magnetic orbitals and the energy gap Δ between the molecular orbitals built from the magnetic orbitals. The tool of the ferromagnetic gladiator is the two-electron exchange integral C between the magnetic orbitals: J = J AF + J F J AF = −2Δ S J F = −2C The concept of magnetic orbital will be specified. The antiferromagnetic gladiator is generally more efficient than his component. This is particularly true when the metallic centers are largely separated (more than ca. 4 A). We shall show how it is possible to tune the efficiency of the tools of the antiferromagnetic gladiator in a given bridging network, by playing on the nature of the terminal ligands. We shall emphasize that the interaction may be very large even when the metallic centers are far away from each other. Some new bridging ligands particularly appropriate for transmitting electronic effects over long distances will be presented. In some quite peculiar cases; it is possible to annihilate the trenchant of the tools of the antiferromagnetic gladiator, by realizing the orthogonality of the magnetic orbitals. J is then positive and the ground state is the state of highest spin multiplicity. This orthogonality can be strict or accidental. Examples corresponding to both situations will be preseted. In this respect, the extraordinary versatility of the azido bridging ligand in stabilizing either the singlet state or the triplet state in copper(II) dimers will be pointed out. In conclusion, the relevance of our work to the study of the polymetallic sites in biological systems will be discussed.