The disagreement between model calculations of ionospheric N2+ concentrations and those measured by the Atmosphere Explorer (AE) satellite has been controverdial for several years. The model calculations of N2+ require an additional loss of the order of ~102 ions cm-3 s-1 at ~250 km in order to account for the observed N2+ concentrations. After analysis of many orbits of AE data we have drawn the conclusion that such a loss rate could only be provided by either (1) enhancing the ionospheric dissociative recombination rate coefficient for N2+ by a factor of 2--3 over its currently accepted value if this is not precluded by laboratory data or (2) enhancing of the charge exchange of N2+ with atomic oxygen. This could be achieved by two mechanisms which either independently or combined can account for the observed N2+ concentrations. The first is the enhanced destruction of N2+ by charge exchange of vibrationally excited N2+ with O (enhancement of the branch to NO+ transfers the problem to NO+). The second mechanism is the accidentally energetically resonant reaction O+(2D) +N2 ⇄ N2+(X)v=5+O. In spite of the valid theoretical arguments against the plausibility of these two mechanisms, without any viable alternatives we feel compelled to present our results. The best agreement between theory and data is obtained when these two mechanisms are combined with the rate coefficients for charge exchange of vibrationally excited N2+ with O increased by a factor of 50 and N2 set equal to 2.5¿10-10 cm3 s-1 and where N2+(X)v<0 is quenched by N2 at a rate of ~5¿10-10 cm3 s-1. An important aspect of the chemical scheme is negligible quenching of O+(2D) by O. We estimate an upper limit for the rate coefficient of this process to be ~5¿10-12 cm3 s-1.