I haven't used MarvinSketch in a while, but isn't it a 2D editor?
Generation of 3D coordinates from 2D files often means trouble. I tried that recently with a 2D BKChem drawing of a trypticene-based ligand for another answer and the result was useless.
I definitely recommend the use of Avogadro as a free 3D editor. Please have a look at their website or the article in Journal of Cheminformatics.
My typical workflow here is:
- switch off auto-adding of hydrogen
- draw the skeleton with the right multiplicity of bonds, but using carbon only
- change the "wrong" carbons to heteroatoms
- let the program add the hydogen atoms
- run a force-field optimization
- manually adjust some dihedral angles
- run the force-field optimization again
- (optional) align a bond with one of the principal axis
- save as CML or MOL
I put a MOL file of etravirine created that way on pastebin. Please give QuteMol a try at it.
I generated a PDB from the molfile mentioned above and ran it through QuteMol. The image below shows that the $\ce{C-Br}$ bond does exist in the PDB file.

Admittedly, checking the PDB file for the respective CONNECT x y
line would have been a (less flamboyant) proof too ;-)