GPS radio occultation observed refractivity is a function of atmospheric temperature and humidity. One-dimensional variational (1DVAR) analysis can be used to retrieve temperature and humidity profiles simultaneously from refractivity data. Profiles derived from 1DVAR method with COSMIC 2011 data occurred in China area are compared with the collocated radiosondes. The background data are ECMWF profiles and the retrieval results show good agreement with radiosondes. The effect of non-ideal gas compressibility on temperature and humidity retrieval errors is focused on in the 1DVAR analysis. Results compared with different months data show that non-ideal gas compressibility has a systematic effect on radio occultation retrieved atmospheric profiles. Including non-ideal gas effect correction the temperature bias to radiosondes is around 0.1 K, while the specific humidity bias is 0.5%. Because radio occultation data can be assimilated into operational numerical weather prediction system without bias correction, taking into consideration non-ideal gas effect is clearly very important. In addition, including this effect in radio occultation retrieval has important impact on climate applications.