This paper presents experimental data that show an over-whelming dependence of system electromagnetic compatibility (EMC) on system cable effects, rather than leakage through cracks, seams, and other box (LRU and WRA) openings. A laboratory experiment compares cable-shield RF leakage from a pigtail-terminated cable with that of one that has been terminated peripherally. Theoretical justification for these empirical observations is presented. That a shielded conductor carrying an RF current i induces its own return current -i on the inside surface of the shield is proved. By confining this return current to the inside of the shield, external emissions are attenuated by the intrinsic shielding effectiveness (which can be quite a large number) of the shield.