We use the ``Via Lactea'' simulation to study the co-evolution of a Milky Way-size LambdaCDM halo and its subhalo population. While most of the host halo mass is accreted over the first 6 Gyr in a series of major mergers, the physical mass distribution [not M_vir(z)] remains practically constant since z=1. The same is true in a large sample of LambdaCDM galaxy halos. Subhalo mass loss peaks between the turnaround and virialization epochs of a given mass shell, and declines afterwards. 97% of the z=1 subhalos have a surviving bound remnant at the present epoch. The retained mass fraction is larger for initially lighter subhalos: satellites with maximum circular velocities Vmax=10 km/s at z=1 have today about 40% of their mass back then. At the first pericenter passage a larger average mass fraction is lost than during each following orbit. Tides remove mass in substructure from the outside in, leading to higher concentrations compared to field halos of the same mass. This effect, combined with the earlier formation epoch of the inner satellites, results in strongly increasing subhalo concentrations towards the Galactic center. We present individual evolutionary tracks and present-day properties of the likely hosts of the dwarf satellites around the Milky Way. The formation histories of ``field halos'' that lie today beyond the Via Lactea host are found to strongly depend on the density of their environment. This is caused by tidal mass loss that affects many field halos on eccentric orbits.

20 pages, 18 figures. Figures 6,7 and 8 corrected in this version, for details see the erratum in ApJ 679, 1680 and http://www.ucolick.org/~diemand/vl/publ/vlevolerr.pdf. Data, movies and images are available at http://www.ucolick.org/~diemand/vl/