Recent experimental measurements in the reflected shock tunnel CUBRC LENS-I facility raise questions about our ability to correctly model oxygen and carbon dioxide recombination. We consider two possible mechanisms involving the electronically excited states of these molecules that may help explain the experimental data. Oxygen has two low-lying electronically excited states, which have long radiative and collisional lifetimes. We postulate that recombination to these states may help explain the apparent errors in predicting the recombination of oxygen. Carbon dioxide has different behavior and has a single excited state just below the dissociation energy. A recent computational chemistry study shows that CO2 recombines to this state and then relaxes to the ground electronic state. We propose a simple model to represent the effect of this intermediate state in the recombination process. Preliminary simulations show that this model may help explain part of the puzzling data.