Hi Aandy,
old hat and unreliable and out of date
I'm sorry Aandy, but I completely disagree with you on all counts. I am rather stunned by your answer really.
There are more problems with your NTC implementation by far. It remains in series with your AC line and will vary it's resistance depending on current flow. A relay to short that mess out is a good idea, by what is wrong with a properly sized resistor instead? I'm sure that may have been where you experienced your earlier problems.
When done properly, a resistor shorted out as Andrew has described will not create hot parts. There is an added plus in that the relay can be made to drop out under fault conditions, causing the resistor to overheat and go open. Many other things hot and smoking at this point so we might as well limit the damage. Keeps the fuse jockeys from destroying the gear.

Seen more than my share of those clowns.
-Chris
Edit: The size of the amplifier has no bearing on the method. You could do this on a 2 wpc amplifier (but why would you?).