The nature of the M–N and N–N bonding and the activation of co-ordinated N₂ towards protonation has been investigated theoretically using the extended Hückel molecular orbital (EHMO) method. Charges on the N₂ atoms and the M–N and N–N overlap populations for the model systems trans-[V(PH₃)₄(N₂)₂]^–, [Fe(PH₃)₄H(N₂)]⁺ and [Fe(PH₃)₄(N₂)][the actual phosphine ligand being 1,2-bis(dimethylphosphino)ethane or its diethyl derivative] show an excellent correlation with experiment. The more reactive N₂ group of the vanadium complex has a much larger negative charge while both the V–N and N–N bonds in [V(PH₃)₄(N₂)₂]^– are computed and observed to be relatively weak with the corresponding bonds in the Fe compounds being relatively strong. Both theory and experiment therefore demonstrate that the simple Dewar–Chatt–Duncanson (DCD) picture is not applicable since it predicts a strong M–N bond should be associated with a weak N–N bond and vice versa. For these first-row transition-metal complexes, the M–N bond is σ dominated while the N–N bond is π dominated and there is no synergic correlation between them. In contrast, EHMO calculations for [Ru(NH₃)₅(N₂)]²⁺ and [Os(NH₃)₅(N₂)]²⁺ correlate both with the experimental IR vibrational data and with the predictions of the DCD model. For these complexes, an increasing M–N interaction is associated with a decreasing N–N bond strength. The factors which lead to the departure away from the DCD picture for the vanadium and iron species are discussed in terms of the metal charges and co-ordination numbers.