I have at the outset to say that I approach discussion of power metering in the context of cycling performance and training with something of a pinch of salt. For many years, I've used heart rate monitors (most recently a Polar S720i wristwatch and bike computer unit) to monitor exertion and effort during my training sessions and in racing. I have a pretty good understanding of where my training levels are (I am still using Pete Read's five levels, for this 0 to 4), and can judge not only where my limits are when racing on that basis, but also can judge when I'm not completely recovered and shouldn't train. This has actually worked pretty well, probably most notably in 2001-3. In recent years I've found that my working life has begun to limit opportunities for training (hopefully addressed now by moving to train before going to work!), and I've been seeking ways to better monitor progress.
In planning my 2009-2010 winter training programme, I'm a bit uncertain as to where power monitoring will fit in (though perhaps interested readers might leave suggestions via the commenting system). I've always felt that the advantage of HRM-based training is that you're getting a readout of physiological response to a training load, which seems to me to be rather more important that knowing what power one's generated. This is, I confess, an element of novelty spurring me on to train more seriously!
I recently bought a Polar CS600X bike computer/HRM, with an associated Power meter system - here I present an initial review of the unit. I later bought the add-on Polar G3 GPS unit for use with it - I'll review this separately as I haven't had a chance to use it at the time of writing.