One possibility came to mind when reading the most recent report (from Swedish group). The device output about 4 KW (thermal) and used a (approximately) 400 watt auxiliary electrical heater coil, with the 400 watt auxiliary input power measured by a calibrated voltmeter and ammeter.
What was not clear was the power source for the auxiliary heater. Depending upon the type of voltmeter and ammeter, it would be possible to use a pulse waveform with a high RMS (heating value) and low average power (as measured by average reading voltmeter and ammeter). So, the "auxiliary" heating coil could be operating at 4KW, but measured at 400 watts, thus providing a fake result.
To detect that sort of fake, it would be useful to continuously observe the auxiliary heater power input waveform with an oscilloscope and also to use calibrated RMS-sensing voltmeter and ammeters on the auxiliary heater power leads.
Another way of sneaking power into the device would be to use the water supply pipes as an electrical heating circuit. This would be detectable with a clip on DC-AC ammeter presumably. (I think plastic or rubber tubing is used from the photos I saw, which would make it more difficult to run power through the pipe, although the tubing could have conductors running through it. Or, one might even use the water as a conductor, if it has impurities.)
Not saying this is a fake, just responding to how one might go about faking the results.
Jack
If you look at the videos of the demonstrations, you will see that there are two temperature controllers active, one for the internal heater and one for the auxiliary heater. Power for both is supplied by a single cable from the 220V wall plug, which is where the power is measured with a clamp-on ammeter (voltage was also measured at that point). So the power measurements at the wall are indicative of total power drawn. No matter what the waveform out of the "blue box" that contains the temperature controllers, the wall current draw will measure total power.
Since some of the demos were held at the University of Bologna (and not Rossi's factory/R&D facility), the likelihood of "jimmying" the wall power is, I think, pretty miniscule.
But if you're interested, there is a 61 page document floating around the web which purports to have an extensive examination of all the possible ways to "fraudilize" the demos. I think there is a link on the Yahoo H-Ni Fusion group:
http://tech.groups.yahoo.com/group/H-Ni_Fusion/