Yes the calculation of F000 must include all atoms in the
crystal cell whether you see them or not.
For disordered solvent the simplest method is to calculate
the volume outside the known structure and then use the expected
solvent density for the crystal mother liquor (i.e. it would
be about 0.334e/A**3 if you had pure water, usually more
for most crystals since you may some salts etc).
Also, don't forget about the hydrogens on the protein,
which probably aren't in your model either.
Another slightly tricky point in accurate calculations is that
real-space models of the protein density tend to be a little
low on electron count since you lose the electron density
in the Gaussian tail outside the atomic radius when you
set-up the model.
Product Development Manager, Macromolecular Crystallography
Molecular Simulations Inc