Xiankai! Haven't seen you for the longest time..
Anyways, according to VSEPR there is a single bonding pair in the C=O bond. I believe this is because the pi bond is parellel (collinear and superimposed around) the sigma bond, and so it doesn't take up the space that other bonding and lone pairs would. A triple bond would be handled the same...as a single bonding pair. So, VSEPR indeed predicts a bond angle of 120° for all of the bonds.
It is known, however, that the double bond has a greater negative-charge-density than a single bond (equivalent to the C-Cl bonds) would. For this reason, the ACTUAL bond lengths in phosgene are slightly different than 120.0°. I found a source citing ~113° for the Cl-C-Cl bonds and ~123° for each O=C-Cl angle. For incredible accurate calculations, a molecular orbital calculation can be done with Gaussian or a myriad of other molecule-sketching programs.
I took my data from
NIST Data Gateway, which is done with a particular method. For HUNDREDS of predictions, you can always go to
http://srdata.nist.gov/cccbdb/geom1.asp.
Best of luck!
Edited April 17, 2006 by lemonoman: The links I gave were crap. One linked back to the post and the other was a bad .ASP. I corrected them both.