A viable technique for remote sensing of the atomic oxygen density and translational temperature profiles above 80 km altitude, where oxygen atoms are involved in important chemical, collisional and radiative cooling processes, would have enormous value. The two profiles can, in principle, be recovered simultaneously from limb scan data obtained near wavelength 147 μm and/or 63 μm, corresponding to the intermultiplet transitions in the ground electronic state of atomic oxygen. Part I of this paper (1989, Planet. Space Sci. 37, 1333) considered an approach in which the limb radiance data represent just one of the lines observed at high spectral resolution; recovery of the temperature relies mainly on Doppler broadening. Here, in Part II, we evaluate feasibility of the experimental concept when the data are a pair of limb radiance profiles corresponding to the two unresolved OI lines; temperature retrieval relies mostly on the difference in upper state populations. O-atom density and translational temperature profiles were retrieved from noise-contaminated synthetic data by an inversion procedure that uses both the onion peeling and global-fit methods. The latter allows solutions to be obtained despite the occurrence of a singular Jacobian, typically at 200 km tangent height. It is concluded that stable profiles for the altitude range 90–300 km can be recovered when the noise-equvalent radiances in the two channels are no greater than roughly 10 −13 W cm −2 sr −1. We estimate that this level of sensitivity could be achieved in a small space-borne cryogenic sensor based on non-scanning Fabry-Perot etalons. The spectrally-resolved approach described in Part I required a considerably larger, more advanced sensor.