Abstract: In 2012, the artificial meteorological observation is replaced by auto weather station in China. More than 2300 surface weather stations begin to use automatic weather observation system, and only 8 national reference climatological stations retain long-term manual observation. Great changes have taken place in observational rules and methods of automatic weather stations comparing with manual observations. Inhomogeneity of data series may be caused by many causes and the change of observational instrument is an important one, and therefore comprehensive comparisons are demanded to gain understanding of their differences.In order to determine the comparability of automatically-observed and manually-observed relative humidity and find out differences between them, the integral point observations from 2007 to 2013 of 8 national reference climatological stations which retain long-term manual observation are analyzed. Analysis results show that the average relative humidity observed by automatic way is from-5.69% to-0.1% lower than that by manual way, and the standard deviation is 2.02%-4.71%. As the humidity rises, the system error of the two increases. Differences between automatically-observed and manually-observed relative humidity are most significant in summer, and related to the wind speed. The difference is obvious under low speed conditions and it decreases gradually as the wind speed increases. The observation of relative humidity is also influenced by the ambient temperature. There is no significant time difference between two types of the hourly observation. The daily variation of differences between automatically-observed and manually-observed relative humidity have a rising trend in the early morning when the relative humidity is higher, especially in the high humid area, and have a dropping trend in the afternoon when the relative humidity is low.The relative humidity data of 8 national reference climatological stations observed by inner screen psychrometer and the hair hygrometer which have some deviations between them, and there are some deficiencies in the measurement performance of humicap too. Besides, the background information of the instruments are not fully known. For all of these, the conclusion has some limitations, although some problems can be explained. It is recommended that the maintenance and calibration of humicap should be carried out regularly, and the performance of the humicap at low temperature should be further tested.