It's overstated to say that who's drawn for the en banc panel invariably matters. It's not too much, however, to say that it's often very important what the wheel brings. Here's a good example.
It's an ineffective assistance of counsel habeas case that gets dismissed by the district court and, back in April of 2008, is unanimously affirmed by the panel, which consists of Judges Beezer, Trott and Randy Smith.
The case then gets taken en banc. Recall that Democratic appointees are more numerous on the 9th Circuit than Republican appointees: the split is 16-11 (or 15-12 if you count Judge Tallman, a Republican appointed by Clinton as part of the Washington deal, where he's most accurately counted). But Judges Trott and Beezer are eligible to be part of the draw if they want since they were on the panel. As it turns out, none of the panel members get drawn, but the draw still results in Republican appointees outnumbering Democratic appointees on the en banc panel 6-5. You gotta like that if you're the government.
But here's the rub. Not everything's rock solid. Votes aren't ordained in stone, nor do they depend entirely on who appointed you or your political beliefs. Sure, there are pretty certain votes in en banc criminal cases, but there's a decent middle -- a fair number of reasonable people who can be swayed.
So, as it turns out, it's indeed a close case: the final vote is 7-4. But not necessarily the way you might expect. Judge Reinhardt writes the en banc opinion, and Judge Bybee authors the dissent. Which pretty much gives away the ending. Judge Reinhardt gets all the Democratic appointees on his side (Judges Silverman, Wardlaw, Fisher, and Paez), and Judge Bybee gets the hard-core conservatives on his (Judges O'Scannlain and Ikuta, plus Judge Kleinfeld).
But Chief Judge Kozinski and Judge Milan Smith join Judge Reinhardt. Which makes for a solid flip. There's your dispositive "middle" in this one. A pretty darn conservative middle, I might add: particularly in criminal non-civil liberties cases. But the dispositive middle it is.
I can promise you the result would have been different if, say, the draw had picked the other Judge Smith. Or if the Chief Judge wasn't invariably selected and instead Judge Rymer took that slot (or Judge Bea was Chief). So sometimes slight differences in draws -- or in procedures -- can make a difference. And, sometimes, you can't necessarily tell the result from the draw. Even if you think you can. Sometimes you've got to actually read the facts and see if this is an especially persuasive case in which some of the less locked-in judges on the Ninth Circuit might come out the way the case factually leans. This is a very good example.
P.S. - The breakdown also gives you a decent hint as to what'll happen to this one at the certiorari stage as well.