I did a study, which I last updated two years ago about how often the BCS "works". I went over every season since 1936. I won't repost all of that, just the introduction and the conclusion:
The BCS system would have “worked”, (produced a championship game between two teams that had better records than any team that finished in the top 8), 18 times in 74 years, counting 1944. That’s 24.3% of the time. The other 75.7%, somebody with a good enough schedule to finish in the top 8 and who had as good a regular season record as at least one of the teams in the championship game would have watched it on TV-or did. (Of course some of those teams lost or would have lost in bowl games before the title game and others lost to a team in the game, so their “case” would have been weakened.) The importance of this is that, if the BCS system isn’t expanded into a more comprehensive playoff, we can expect there to be a significant controversy, at least going into the bowls, ¾ of the time. The BCS is a .250 hitter.
2011 Update: 18 times in 76 years .236. If it had been a four team playoff, as is now being proposed, There would have been 23 years where the #5 team had as good as record as the teams in the tournament. That’s 30.3% of the time there would still be a problem.