The basic structure of Rayleigh-Raman-Mie (RRM) lidar as well as the principle of temperature retrieval algorithm using Raman scattering are introduced. Background correction, slide average and wavelet transform are employed to denoise the Raman backscattering signal, and then the effect of aerosols to temperature profile is analyzed. Using the algorithm described, temperature profiles above Nanjing from 5 km to 18 km during 1853 to 1935 on November 19, 2010 are retrieved. The measured results showed that the temperature variation at altitude of 5.5 km is less than 2 K; The entire month of observation data during November 2010 are analyzed and processed, average temperature profiles of three ten days are obtained. At altitude of 10 km, temperature of first ten days is 4 K higher than that of the third ten days. With winter approaching, the temperature lapse rate has significantly increased in low-altitude section. Monthly mean temperature in the range of 5~10 km is 4 K lower than the model value, and the two profiles are almost parallel which showed temperatures at most altitude below 10 km are 4 K lower than that of model.