Anyone know why the 3sigma orbital in CO has such a large coefficient on Carbon. I would expect it on Oxygen because Oxygen is more electronegative and because the 3sigma orbital is somewhat low in energy. I suppose it's just one of those gray lines.

To make up for the large coefficient on oxygen in the NHOMO, the HOMO has to have a small coefficient on oxygen.  

Is that right? I'm not so sure about NHOMO affecting a HOMO of a different atomic orbital set, though it's true that  to make up for a large coefficient in a bonding MO, the antibonding version gets a small coefficient.

My take is that it's mostly due to some mixing of s character from the carbon  with the pz orbitals of C (sp hybrid) and pz of O to form the sigma bond MO.
The O exercises its electronegativity mostly through the pi bonds as movies observed.

Interestingly, the large HOMO C coefficient  explains why CO bonds through the C in metal carbonyls.

Doh!  My bad.  You guys are right.  The polarization of the pi orbitals won't affect the coefficients in the sigma series.

I checked in my inorganic textbook and it seems to suggest essentially what Demotivator said.  The 3-sigma MO arises from the p_z AO on oxygen and the 2s and 2p_z AOs on carbon.  

It seems that the 1-sigma MO represents a lone pair on oxygen (essentially non-bonding) and the 2-sigma MO represents the sigma bond between C and O.  The 1-pi MO only involves p-orbitals on the C and the O, so it won't affect the 3-sigma orbital.

Any population of the 3-sigma orbital would be a bonding interaction with respect to the p AOs involved, but anti-bonding with respect to the s AO on carbon (the bonding sigma interaction is accounted for in the 2-sigma MO).  Since the s AO on carbon is lower energy than both the p AOs on carbon and on oxygen, the lowest energy MO will be primarily composed of the s AO on carbon, suggestive of the localization of those electrons on carbon.

Thanks for catching my error guys.