After being told by funkybulb that I can calculate the total power draw of an HID ballast system by multiplying the input current by the line voltage and getting unrealistically low power draw readings with my kill-a-watt, I am starting to have doubts about its accuracy. I am wondering if a Kill-A-watt meter is really that accurate in measuring the total power draw of a lighting system?

From my experience, they're pretty accurate, however, they can become inaccurate if the unit overheats and damages the current shunt. So if the unit appears to have overheated around the outlet, then that could be your issue. If not, then there could be an issue with the calculation, or the Kill-A-Watt could have some internal damage that's not visible from the outside.

BTW, by current, do you mean Watts, Amps, or Volt-amps?

By multiplying rms current and voltage you get just an apparent power, not the real power.
In order to get the real power, you need to multiply the apparent power by the power factor. But the practical problem is, the power factor uses to be unknown.

The Kill-a-Watt (and similar real power meters) does follow all the instant power flows (instant power is instant voltage multiplied by the instant current at the same moment; not the average nor rms, but really a voltage and current sample; it then acquire few 100's of such samples so get 100's samples of the instant power each mains period) and by averaging them it get the exact real power (averaging all the instant power samples over some whole number of mains periods).
In this way it is able to really follow all real power, regardless what is the cause of the nonunity power factor.
There is just one caveat: Typical implementation uses only AC coupling for both current and voltage signals (in order to avoid the need for extremely low offset signal processing and to allow the use of a current transformer to pick up the current signal so the thing remains robust against overloads or so), so does not count for power transfered by DC components. But because that could happen only when both current and voltage have a significant DC component, in real life it is not any problem (the voltage can not have DC component bigger than some 100's mV, so the error from neglecting the DC component is below few %, so within the acvuracy spec of the instrument). But it is the reason, why it can not be used for DC circuits.