The dissociation and isomerization stability of the linear cyanogen isomer CCNN, a formally charge-transfer species comprising C-2(-) and N-2(+), is theoretically investigated by means of ab initio methods, including the Hartree-Fock (HF), the Moller-Plesset perturbation theory from second through fourth order (MP2, MP4SDQ, MP4SDTQ), the configuration interaction with singles and doubles (CISD), the quadratic configuration interaction with singles and doubles (QCISD) as well as with triples [QCISD(T)], and the density functional theory (DFT) including Beck's three parameter hybrid methods with the Lee-Yang-Parr correlation functional (B3LYP) and with the Perdew-Wang 91 correlation functional (B3PW91) methods. At the QCISD(T)/6-311G(3df)//QCISD(T)/6-311G(d) level with QCISD/6-311G(d) zero-point vibrational energy (ZPVE) correction, the barriers from CCNN to the NNC three-membered ring structure with exocyclic C-C bonding and to the dissociation products C-2 and N-2 are predicted to be 42.1 and 51.8 kcal/mol, respectively. The QCISD(T)/6-311G(3df) potential energy surface of C2N2 indicates that CCNN may be kinetically more stable than the other two well-studied isomers CNCN and CNNC. Thus, although CCNN is thermodynamically less favorable due to its high energy, it is still experimentally observable. Moreover, a barrier of 5.5 kcal/mol is predicted for the reverse association reaction between C-2 and N-2 to produce CCNN at the QCISD(T)/6-311G(3df)//QCISD(T)/6-311G(d)+ZPVE level, while both B3LYP and B3PW91 methods erroneously predict a barrierless association process. Finally, the possible strategy for the formation of CCNN in laboratory and in interstellar space is discussed in detail. The calculated results may provide a useful guide for future laboratory and interstellar identification of the last kinetically stable isomer of cyanogen, CCNN.