We evaluate the degradation of the accuracy of the component separation between the cosmic microwave background (CMB) and foreground components caused by neglect of absorption of the CMB's monopole component by the galactic interstellar matter. The amplitude of the temperature anisotropy caused by the CMB shadow, due to dust components, is about 1 μK. This value is comparable to the required noise level necessary to probe non-Gaussianity studies with upcoming CMB experiments. In addition, the amplitude of the polarization caused by the CMB shadow due to dust is comparable to or larger than the rms value of the CMB B-mode polarization, imprinted by primordial gravitational waves. We show that applying a single-power-law model as the dust spectrum to observed multifrequency data introduces systematic errors, which are comparable to or larger than the required noise level for forthcoming CMB B-mode polarization experiments. Deducing the intrinsic spectrum of dust emission from the submillimeter wave band data reduces the systematic error to below the required noise level. However, this method requires dust temperature measurements with an accuracy of better than a few percent. We conclude that the CMB shadow due to dust must be considered in future CMB missions for achieving their targeted sensitivity.