For some reason, I can't think of a storm that explicitly tracked well off course of model projections (especially within a day or so). At this point in time, the model outputs should become increasingly consistent and refined. If not, then there might be something that is "missing" from the initialization of the pattern. I want to say that Charley made landfall south of most of the model projections last year. There may be numerous other examples, but there aren't many that come off the top of my head.

However, if you live along the west coast of Florida, then a landfall area between Appalachicola and Pensacola might be a big deal. The Appalachicola track would put the west coast of Florida at a greater risk for dangerous conditions...the Pensacola track allows for more breathing room. These two locations aren't extremely far from each other, and it's possible that future model runs may flip-flop between these two lacations. This is why everyone in Florida needs to be watching this storm.

On the flip side, hurricane conditions over the Peninsula (from Dennis) look unlikely at this time.