Why is the observed Western Blot band size different from predicted size?

The predicted M.W. is based on protein sequence analysis; however, some factors might lead to an observed band size that is different from the predicted size. The reasons might include:

1.Post-translational modification (PTM):
  a. Some post-translational modifications might lead to increased protein size, including
      phosphorylation, acetylation, methylation, glycosylation, sumoylation, ubiquitination,
      etc.
  b. Some post-translational modifications might lead to decreased protein size including
      phosphatidylethanolamine conjunction (e.g. LC3-II)
  c. Some proteins may be cleaved to form an active or mature form; this process will
      lead to a decreased protein size (e.g. Notch activation, Caspase activation, etc.)
  d. Some websites provide useful PTM information

   i. HPRD http://www.hprd.org/
   ii. ProSite http://www.expasy.org/prosite/
   iii. ELM http://elm.eu.org/
   iv. CBS data sets http://www.cbs.dtu.dk/databases/
   v. CBS prediction Servers http://www.cbs.dtu.dk/services/
 
2.mRNA splice variants (Isoforms):
Through alternative splicing, one gene can generate different proteins with different M.W. Regulation of alternative splicing depends upon cell type, conditions, etc.

3.Multimerization:
Some proteins could form dimers or multimers, increasing the M.W. This phenomenon usually can be found in reducing gel condition; however, strong interactions may still be seen with higher molecular weight proteins even in denaturing gel.

4.Protein charge:
The observed size could also potentially be influenced by the protein charge

5.Different species:
Different species likely have different protein sequence and PTM, which can lead to a different protein M.W.

Source: http://www.genetex.com/WebPage/Product/PredictedTarget.aspx