Abstract：Taking elevation component time series of GNSS permanent stations as examples, effects of data defect and noise level on results of GNSS time series analyzing are discussed based on both simulation and observation data. The results show that the higher the noise level is, the greater the effect will be. If the days of data defect are the same, the effect of continuous data defect is much greater than the effect of discrete data defect. The interpolation can’t remedy the data defect effectively in a high noise level if days of data defect are more than a certain number of days. On the basis, the request of observation data in different reliability requirements is discussed. It is pointed out when annual and shorter periodic variations are considered and to ensure deviation of linear velocity be less than 10%, for example, when the noise lever is 9mm, then the defect days for continuous data losing should be less than 11 days and for discrete data losing, the defect days should be less than 12 days. Last but not least, an advice is given that in the study of vertical crustal movement using elevation component time series of GNSS CORS, the operation status of GNSS permanent station should be strictly screened.