The zerovalent bis(eta(6)-benzene) f-metal sandwich complexes M(C6H6)(2) (M = La, Ce, Nd, Gd, Tb, Lu, Th, U) were investigated with state-of-the-art quantum chemical ab initio approaches taking into account the effects of electron correlation and relativity. Ground state assignments, optimized metal-ring distances, symmetric metal-ring stretching frequencies, and metal-ring bonding energies are reported. The effects of ring substitution on the metal-ring binding energies are discussed. The complexes of Th and U are predicted to be at least as stable as the corresponding lanthanide systems and form possible synthetic targets. Whereas the lanthanide systems have a 4f(n)e(2g)(3) ground start configuration (n = 1, 3 for Ce, Nd), the corresponding actinide compounds should possess, as a consequence of stronger relativistic effects, a 5f(n-1)e(2g)(4) ground state configuration, with a possible strong admixture of 5f(n-1)a(1g)(2)e(2g)(2) (n = 1, 3 for Th, U). The back-donation from the occupied metal d(+/-2) to the empty pi orbitals of the benzene ligands is Found to be the dominant bonding interaction. Whereas the lanthanide 4f orbitals are essentially localized on the metals and chemically inactive, the actinide 5f shell is partially delocalized and its f(+/-2) components may also take part in the back-donation.