This paper carefully examines the inability of photochemical models to account for the large nitric oxide densities of ~108 cm-3 at ~105 km obtained from IR, UV, and microwave measurements. A detailed and up-to-date photochemical model is constructed that incorporates measured YOHKOH soft X ray fluxes, hot N atom chemistry with an energy dependent thermalization cross section and seven reaction sources, and laboratory-constrained N(2D) yields. The resulting model which has well-constrained chemistry compared to past models fails to generate high enough NO densities in comparison with the most reliable measurements of absolute NO concentrations in the lower thermosphere. The sensitivity of the model results and the known uncertainties in the inputs are used to identify where future efforts should be focused. A deficit remains despite an increase in the vertical mixing rates in the lower thermosphere from the very low Kzz profile used in our calculations and/or an increase in the N(2D) yield from electron impact dissociation of N2 from its nominal value of 0.54 to 0.62. The sensitivity of NO profiles to the nascent energy distributions of the atmospheric sources of suprathermal N atoms is illustrated by including the thermalization of suprathermal N atoms with an updated thermalization cross section. The diurnally averaged NO concentration at 105 km is enhanced by factors of 1.2 and 2.6 when the energy distributions of the N atoms from electron impact dissociation of N2 are chosen with peaks near 0.6 eV or 3--4 eV, but deficits of factors of ~7 and ~3, respectively, remain. There is higher sensitivity to vertical transport than to variations of chemistry within known uncertainties. ¿ 1998 American Geophysical Union