Nick, do you think a glitch like that might be power-supply related? I've only used the low-power modules, but a friend has some Pros and I know he ran into some sort of power issue that was fixed with an external supply. I forget the specifics, I'll check back with him.
I couldn't resist trying to duplicate the problem with my low power S2 modules, but failed miserably, which is to say everything worked fine. I used a modified version of the "client" and "server" sketches posted previously.
I am a hands-on type, and another obvious thing that I didn't realize until I put the hardware together, is that when the "server" sketch prints the received characters, the XBee is of course sending them back to the "client" unit. I was fiddling with two-way communication and got a pretty nasty echo loop going which is what made that light come on for me. So @falven, one more thing to try, yank the jumper from the JP3 header on the "server" shield, to prevent that XBee from echoing back what it receives. Not sure why that would make a difference for you when it's not an issue for me, but what with different shields, S1 vs. S2 and all, might as well try it, it's simple enough.
Here are links to a screen shot and a pic of my hardware setup.
http://img546.imageshack.us/img546/9849/screengx.jpg
http://img4.imageshack.us/img4/4834/hardwareb.jpg