Spencer Ackerman makes an observation that can't be repeated enough:
So I asked Karen Greenberg, director of NYU's Center on Law and Security to help me out here. According to Karen, if you count dismissals by either the judge or the government, the conviction rate in terrorism cases is 87 percent. If you don't, the rate is 95 percent.
That's compared to a 91 percent conviction rate (.pdf) for all criminal cases.
Initially, I asked about this wondering if the stats wouldn't be something like 98 percent or something -- in other words, an indication that maybe we counterintuitively have akangaroo court system for civilian terrorism convictions. But the stats don't really bear it out. If the rate hovers around the 91 percent all-convictions rate, then if we have a problem with the fairness of convictions at all, it's hardly limited to terrorism. Score one for getting the data.
This is so much better in chart form! Using the data from Greenberg's 2010 report that is, which has the number at 598 out of 688 cases (not sure where Spencer gets 400).