well are you testing sources, in terms of their headphone output stages, or are you feeding the line output into something like a headphone amp, receiver, or something similar?
some devices have generally good sound quality, but crap headphone output abilities (most computer soundcards fall into this category, look at something expensive like the X-Fi Elite Pro, great device, the headphone amp is probably equivalent to a $50 mp3 player (at best), while the rest of the card is equivalent to some mid-range dedicated source devices)
granted, my CDP's headphone output stage is better than any other of my source components' headphone output stages, I'll still take a receiver or amp over it any day (you can almost hear it straining to drive the headphones, not like hiss or cracks or anything, but you can tell it wants to open up more than its able to under the load (especially once you feed the fixed (or even using the variable output which is on the same pot and everything) into an actual amplifier)
that'd be my only aversion to using headphones, is that if you're driving the headphone output stage on your source, you aren't getting an accurate image of what the device can do (unless you've got something like Grado's, even still, not guranteed that it won't be placing a bit of un-due strain on the device that using the line outs would avoid), although if you're going from whatever source (lets say a CDP for the sake of discussion) into an amplifier, receiver, integrated amplifier, or headphone amplifier, and connecting whatever to that (speakers, headphones, plasma tweeters, you know, whatever your heart desires) I'd just go with something you know
for example if you've got a pair of speakers that you know intimately, and you know how they'll sound with a given piece, compare with the new piece, see if you like the sound, basically only change one variable
comparing two CDPs (for example) with headphone outputs, via headphones, is changing more than one variable, first, you're comparing their D/A's and transport abilities (thats one or two variables, depending on if your system has a DAC (or if you're adding one)), AND the headphone amplifier circuitry, which is two variables right there
if you add a DAC (as mentioned), that makes it three variables, especially if we're talking different DACs (i.e: if you wanted to compare two really high end transports, lets say Metronome's Kalista and Esoteric's P-03, which both come with outboard D/A units, although if you consider those outboard D/A units to be unique to those systems (i.e: you don't want to try mix and match with the Esoteric D/A on the Metronome transport), and feed only line out, you will still be fine (theres a lot of other variables comparing a pair of players like the Kalista and P-03 introduce that I'm ignoring, because thats really beyond the scopre of what you're asking (I think...))
