Bug in ChemBioDraw.
There was a blog entry on In the Pipeline about a bug in ChemDraw. Actually this has been known for a while (and present in previous versions) but it seems it still has not been fixed in the latest version of ChemBioDraw 13 on the Mac. As you can see in the image below including explicit hydrogens in your structure significantly impacts the calculated LogP. Whilst people don’t often add explicit atoms to phenyl rings, (expect perhaps in SAR studies) they often add them to heteroatoms.
At the moment there is no bug fix and no date set for a fix to any version of ChemBioDraw, the only approach is to avoid adding explicit hydrogens to structures if you want to calculate LogP. I’ve looked at a number of other applications and there seem to be no issues with ChemDoodle, Elemental, Marvin or OpenBabel.