1.25" seems like a lot to me.  With a 30" deep beam, I am assuming that it is taking a substantial load.  That being said, that 'misalignment' could be putting a relatively significant amount of torsion (since W shapes have a poor response to torsion) into your beam, which was probably not accounted for in design.  
This may not be an issue depending on the spacing of your lateral braces, but I would definitely check the beam using an e=0.75" for all loads.
Just to make sure I am thinking of this correctly, I am picturing a beam spanning 30' with beams framing in on one side and the bottom of the supporting beam is 'kicked out' by 1.25" to the side opposite where the beams are framing into it.

MYerges,

Thank you for the response.  It is actually a 120' long PEMB rafter with "Z" purlins running continuous over the top of the rafter.  It is a double slope frame so the eave to ridge distance is 60'.  Flange brace angles run down from the purlins at ~45 degrees and brace the bottom flange of the rafter.  It is possible that the flange braces were bolted to the incorrect pre-punched holes on the purlins thus pulling the bottom flange to the side.  I agree that a torsional analysis would be in order if the beam can not be brought plumb.

No fat in those things.  Is it possible that the bracing was not installed until after the beam buckled?  Maybe there was not enough bracing in the first place?  

I think theoretically if there were enough bracing, the beam could be analyzed by calculating new section properties based on its rotated section, which in this case wouldn't be much different than the non-rotated shape.  So I think adding bracing would be the way to go just to make sure.