The cleaning procedure used to produce the data that we analyze for the search of periodic sources of gravitational waves is based on different steps, which are applied to both time and frequency domain data. We have recently improved the procedure, which now consists of different steps. The use of a cleaned procedure is in principle important, since it is aimed to recover at best the observation time from the data by vetoing only times where disturbances act and not entire data chunks. Clearly, the effect of the procedure depends on the nature of the data, and is thus highly related to the detector characteristics in a particular run. We will here describe the whole cleaning chain, by giving details and examples based on the C7 and WSR10 Virgo runs.