Magnetocrystalline anisotropy (MA) constants are calculated for numerous Ni(m), Co(m), CO(Nz)/Pd(p)/Co(m), Pd(p)/Co(m)/Pd(p) (001) and (110) fee slabs using a realistic tight-binding model. Perpendicular MA for an (001) fee Co/Pd interface is predicted in agreement with experimental data and other theoretical calculations. MA constants are broken down into contributions coming from various k points in the two-dimensional Brillouin zone or from various atomic layers in a slab. it is visualized that MA originates mostly from the regions around surfaces and interfaces though contributions to MA come also from inner regions of a slab. The empirical rule of adding MA constants coming from different surfaces and interfaces present in the system is examined against the theoretical results obtained. The validity of the state-tracking technique used recently to improve convergence in MA calculations is also discussed.