In the United States, the voltage of typical wall outlets is 120 Vac (RMS). Maximum current draw is 15 A or 20 A. Is this "maximum" expressed in terms of RMS or peak? The "maximum" seems to imply peak.
Extension to this question: If I plug multiple devices into a power strip with a 15 A fuse, should I be worried about the total peak current or total RMS current of the devices?
Like Brian said, it's measured in RMS.
For the extension: RMS again. For fast-blow fuses, the current rating is the maximum current guaranteed at which the fuse will not blow. But the fuse won't immediately blow at 15.1A -- the time vs. current graphs are provided for most fuses. Here's an example:
As we can see, a 15A fuse is guaranteed not to blow at 15A. As the current increases, the time it takes for the fuse to blow decreases. A 15A fuse may take a minute to blow at 20A, but 0.3 seconds to blow at 30A.
Even though your 15Arms current is actually over 21A peak, the time that it goes above the 15A rating (~4ms per half cycle) is not sufficient to blow the fuse. Due to the fuse's thermal inertia you may want to derate the curve a bit, but at 15Arms you should be fine.
All ratings in the NEC are in RMS. This is explicitly set out for voltage ratings in the Article 100 definition of voltage:
Voltage (of a circuit). The greatest root-mean-square (rms) (effective) difference of potential between any two conductors of the circuit concerned.
but is implied for current ratings in the Code. This is because the current-carrying capacity of a wire is determined by what temperature the wire can reach without damaging its insulation, terminations, or surroundings, and RMS currents are used when the heating value of a given current is at stake. Likewise, fuse and circuit breaker ratings are RMS currents as well -- fuses operate on heat, and so do the thermal trips on circuit breakers.
Most fuses and breakers rated under the UL scheme, however, are not intended to withstand 100% of their rated current continuously, and this is reflected by the 125% derates for continuous loads found in several places in the NEC. (This differs from the IEC fuse rating scheme, which does allow fuses to be run at 100% of their rating continuously.)
It's RMS, but not for the extra reasons given in the other answers. It's nothing to do with the power calculations, or the time lag of fuses, or the temperatures of conductors.
Of the several options (peak, RMS, rectified DC average etc), RMS is the one that's been chosen.
Once a standard way of making the measurement has been chosen, all the other ratings can be adjusted to suit. For instance, if we measured voltage and current as peak, then we would simply rate our kettles at twice their present wattage as peak watts, and our cables at 1.414 their present current as peak amps. No domestic user 'knows' or even cares whether a 2kW kettle is 'really' 2kW, what they need to know is how fast it will boil, and can it be connected to this socket? Experience and the ratings will tell them those. The measurement scientists will understand what's going on, and make the appropriate calculations.
As it happens, RMS is convenient for the measurement scientists, it does reduce the number of conversion factors. But convenience rarely has a look-in around electricity supply technology, look at the conversion factors needed to use AWG rather than mm2 for cables?
