The recently observed upper limits to the beryllium abundances in Population II stars are much lower than Population I detections. We show that this difference reflects an intrinsic difference in the initial abundances and is not caused by different degrees of depletion driven by stellar evolution processes from similar initial abundances. We have constructed evolutionary sequences of models from the early pre-main sequence to beyond the turnoff that correspond to halo dwarfs with [Fe/H] = -1.3, -2.3, and -3.3. We have considered standard, diffusive, and rotational mechanisms to estimate a maximal possible beryllium depletion. We find that halo star models in the range 6000 ≥ Teff ≥ 5000 K might be rotationally depleted by a factor of 1.5-2, and that the total depletion should be no more than (conservatively) a factor of 3. Implications for cosmology, cosmic-ray theory, and Galactic chemical evolution are discussed.