Ab initio molecular orbital theory has been used to calculate accurate enthalpies of formation and adiabatic electron affinities or ionization potentials for N-3, N-3(-), N-5(+), and N-5(-) from total atomization energies. The calculated heats of formation of the gas-phase molecules/ions at 0 K are DeltaH(f)(N-3((2)Pi)) = 109.2, DeltaH(f)(N-3(-)((1)Sigma(+))) = 47.4, DeltaH(f)(N-5(-)((1)A(1)')) = 62.3, and DeltaH(f)(N-5(+)((1)A(1))) = 353.3 kcal/mol with an estimated error bar of +/-1 kcal/mol. For comparison purposes, the error in the calculated bond energy for N-2 is 0.72 kcal/mol. Born-Haber cycle calculations, using estimated lattice energies and the adiabatic ionization potentials of the anions and electron affinities of the cations, enable reliable stability predictions for the hypothetical N5+N3- and N5+N5- salts. The calculations show that neither salt can be stabilized and that both should decompose spontaneously into N-3 radicals and N-2. This conclusion was experimentally confirmed for the N5+N3- salt by low-temperature metathetical reactions between N5SbF6 and alkali metal azides in different solvents, resulting in violent reactions with spontaneous nitrogen evolution. It is emphasized that one needs to use adiabatic ionization potentials and electron affinities instead of vertical potentials and affinities for salt stability predictions when the formed radicals are not vibrationally stable. This is the case for the N-5 radicals where the energy difference between vertical and adiabatic potentials amounts to about 100 kcal/mol per N-5.