We investigate the LHC discovery potential for electroweak scale heavy neutrino singlets (seesaw I), scalar triplets (seesaw II) and fermion triplets (seesaw III). For seesaw I we consider a heavy Majorana neutrino coupling to the electron or muon. For seesaw II we concentrate on the likely scenario where the new scalars decay to two leptons. For seesaw III we restrict ourselves to heavy Majorana fermion triplets decaying to light leptons plus gauge or Higgs bosons, which are dominant except for unnaturally small mixings. The possible signals are classified in terms of the charged lepton multiplicity, studying nine different final states ranging from one to six charged leptons. Using a fast detector simulation of signals and backgrounds, it is found that the trilepton channel l+- l+- l-+ is by far the best one for scalar triplet discovery, and for fermion triplets it is as good as the like-sign dilepton channel l+- l+-. For heavy neutrinos with a mass O(100) GeV, this trilepton channel is also better than the usually studied like-sign dilepton mode. In addition to evaluating the discovery potential, we make special emphasis on the discrimination among seesaw models if a positive signal is observed. This could be accomplished not only by searching for signals in different final states, but also by reconstructing the mass and determining the charge of the new resonances, which is possible in several cases. For high luminosities, further evidence is provided by the analysis of the production angular distributions in the cleanest channels with three or four leptons.