help with ct of reference gene - (Sep/20/2010 )

Can anyone tell me the acceptable difference in average ct number between a reference gene and an experimental gene?
For eg., When ref gene A has ct of about 15, reference gene B about 22 and my experimental is 25
WHen normalised to A: 25/15 = 1.66
When normalised to B: 25/22 = 1.33
Is the data more reliable when the ratio is closer to 1?

Thanks!

Usually it's better to use the reference gene with a Ct closer to the target, but important thing is also a quality (stability) of the reference gene, not only the level of its expression. If both seem stable, then use the one with 22.