Two new techniques have been developed to estimate the densities of N2, O2, and O from Rayleigh lidar backscatter photocount profiles and from independent temperature determinations. Both techniques involve solving initial value problems for the N2, O2, and O densities. These initial value problems are derived from the lidar equation, the ideal gas law, and the assumption of hydrostatic equilibrium. Their solutions are valid in regions where the Rayleigh lidar backscatter can be fully isolated from other scattering sources, e.g., aerosol scattering and where independent temperature determinations can been made. Estimates of the N2 and O2 density profiles in the upper mesosphere and lower thermosphere have been performed on three nights using simultaneous sodium-resonance-fluorescence lidar measurements to independently determine the temperature profile. The available measurements are of insufficient quality to retrieve O density profiles. The resulting N2 and O2 density retrievals appear reasonable when compared with previous determinations from sounding rockets and give hope that this method will allow routine ground-based measurements of major constituent density in the middle atmosphere. A detailed error analysis is described which shows that the primary source of random errors is the Rayleigh photocounts, while the most important systematic errors are uncertainties in the N2 and O2 Rayleigh backscatter cross sections and in the normalization of the lidar density profiles in the upper stratosphere and lower mesosphere. ¿ 2001 American Geophysical Union