We investigate the formation of the stellar halos of four simulated disk galaxies using high resolution, cosmological SPH + N-Body simulations. These simulations include a self-consistent treatment of all the major physical processes involved in galaxy formation. The simulated galaxies presented here each have a total mass of ~10^12 M_sun, but span a range of merger histories. These simulations allow us to study the competing importance of in-situ star formation (stars formed in the primary galaxy) and accretion of stars from subhalos in the building of stellar halos in a LambdaCDM universe. All four simulated galaxies are surrounded by a stellar halo, whose inner regions (r < 20 kpc) contain both accreted stars, and an in-situ stellar population. The outer regions of the galaxies halos were assembled through pure accretion and disruption of satellites. Most of the in-situ halo stars formed at high redshift out of smoothly accreted cold gas in the inner 1 kpc of the galaxies potential wells, possibly as part of their primordial disks. These stars were displaced from their central locations into the halos through a succession of major mergers. We find that the two galaxies with recently quiescent merger histories have a higher fraction of in-situ stars (~20-50%) in their inner halos than the two galaxies with many recent mergers (~5-10% in-situ fraction). Observational studies concentrating on stellar populations in the inner halo of the Milky Way will be the most affected by the presence of in-situ stars with halo kinematics, as we find that their existence in the inner few tens of kpc is a generic feature of galaxy formation.