I am used to retention time decreasing when temp increases. I am testing a growth factor by WCX in 25mM MES at 40C on an HP1100 detection by UV at 280nm.

OK, this is a blind shot at it, but if your growth factor is a protein, maybe the temperature shift is causing a change in conformation that will change the way the molecule interacts with the column. Another thought (though I would think it unlikely) is that a change in temperature will give a small change in pH and maybe that has something to do with it. (?)

What kind of temperature shift are we talking about here? 5 deg C? 50 deg C?

One wierd thing that I saw long ago was a system that had a small leak at the column inlet that apparently was slightly exacerbated by higher temperatures, and exhibited a similar phenomenon to what you describe. The leak was hard to catch, too, as the mobile phase was close to being neat methanol (with no non-volatiles) and the column temp was pretty high, causing the leaked mp to evaporate quickly and without any residue.

I would think though, that you'd be able to spot a leak prett easily with the MP / temp conditions you describe, but I thought I'd put it out there anyway.

I think MES has a change in pKa of about -0.01/C. So if you heat your buffer by 10 degrees C you are going to shift your pH lower. At a lower pH your protein will have a more positive net charge. Since your separation is by cation exchange, I would expect greater retention. I think this should be the general trend for cation exchange of proteins.

Adsorption chromatography (hydrophobic interaction) of proteins is said to be entropy driven, thus strongly temp. dependent and usually associated with an increase of binding parallel to a rise in temp. Now, the molecular explanation of the role of entropy is via a release of H2O during the adsorption process, that is, an increase in the disorder of water. Now one can imagine a similar release of water in ion exchange?