A question may be asked - this EMI amplification happens only at lower frequencies - does it matter?  Yes, it does.  When equipment is tested in the EMC lab for compliance, the cables are relatively short, their own parasitic capacitance and inductance are quite low.  How long do you think are your power cables? Don't just think of that cable with the plug coming from your equipment - think of the network of all power and ground cables in your facility, all with their inductances, capacitances, resonances, resistance, skin effect, and the rest.  When a signal with a broad spectrum such as sharp pulse (shown in the second chart), encounters cable maze, the shape of the waveform changes.  When equipment goes through EMC tests, the low frequency signals are simply not present - nothing to worry about.  But once you install this equipment in your facility, there aren't many broad-spectrum signals anymore but quite a lot of the low-end-of-the-spectrum signals - exactly in the region where regular filters amplify signals rather than suppress them.  This is illustrated in our third figure that perhaps requires not much explanation by now - it all corresponds to the filters' own specifications.  In most facilities there isn't much of energy in signals above 2MHz - for the most effective noise suppression this is the frequency range that is important for actual installations.