A coworker just pointed me to this pretty scary article:

Rolling Stone : Was the 2004 Election Stolen?

My first reaction was that it was just going to be liberally biased FUD - but the author has clearly done his homework and there are some pretty freaky statistics and studies quoted in there. For example, regarding exit polls (that is, asking people "what did you vote" directly after they voted... usually an extremely exact prediction of voting results) :

The evidence is especially strong in Ohio. In January, a team of mathematicians from the National Election Data Archive, a nonpartisan watchdog group, compared the state's exit polls against the certified vote count in each of the forty-nine precincts polled by Edison/Mitofsky. In twenty-two of those precincts -- nearly half of those polled -- they discovered results that differed widely from the official tally. Once again -- against all odds -- the widespread discrepancies were stacked massively in Bush's favor: In only two of the suspect twenty-two precincts did the disparity benefit Kerry. The wildest discrepancy came from the precinct Mitofsky numbered ''27,'' in order to protect the anonymity of those surveyed. According to the exit poll, Kerry should have received sixty-seven percent of the vote in this precinct. Yet the certified tally gave him only thirty-eight percent. The statistical odds against such a variance are just shy of one in 3 billion.