Several early auroral observations of enhanced radiation in the 1.27-&mgr; band of O2 were suggested as being due to the tapping of a reservoir of vibrational energy. The cross section for the process O2(3&Sgr;-g(&ngr; +3)) e-→O2(1Δg) (&ngr;=O) is here estimated by elementary compound state theory and found to be much larger than that predicted by the first Born approximation formula but still too small for this mechanism to be a dominant feature in 1.27-&mgr; radiation in aurorae. Appreciable populations of higher vibrational states of O2(1Δg) are predicted by compound state theory, and these can be a significant feature of auroral infrared radiation before they are relaxed by further collisions.