The resulting shift between a non glycosylated protein and a glycosylated protein would depend on the number of glycosylation sites present on your protein. In addition glycosylation may be hetrogenous. So the mass can change dramatically by several thousand daltons depending on the extent of protein glycosylation
Also consider that the sugar side chains added in the process are fairly big and don't present interaction with the SDS in the same expected way as amino-acids do. The sugars make it difficult for the protein to interact with the SDS and in certain occasions, terminal sugars may present charge, altering the mobility of the proteins in the gel giving different shifts or even smears in the gel.
thank you very much Marc, this does shed a light on my problem... although , I seem to have 3 probable sites according to the prediction software online... how much of a shift can be expected?
If your protein was originated In vivo, the glycosylation grade will be variable, and you will see more kind of a smear than a band in your gel . This smear will be originated by a mixture of proteins glycosylated with different patterns (2 sites glycosylated one no, 3 glycosylates, etc). In addition some of the glycosilated species will have a Mw much higher of the theoretical Mw of the a-glycosilated specie, it will depend on the glycosilation grade obtained and the glycosyl specie added.