The probabilities relate the chance that the center of the storm will pass within 65 miles of the region within the next 72hr. Any storm more than 2 days out is going to have a wide range of probabilities at the later periods, both in number and over a large area. Or, in other words, even for an expected perfect track forecast at 72hr, you would still see areas up to 200mi on either side of the cone listed in the probabilities.

Note that for Rita, over 3 days from landfall, the probabilities are going to be rather low and weighted toward areas closer to the coastline along the Northern Gulf of Mexico -- like New Orleans -- and not to where the actual track takes the storm. Climatology, avg. track error, model spread, and a number of other factors go into determining these probabilities and, yes, you will see some weird results from time to time as a result of some of the factors. I wouldn't pay much attention to a 1% value, though -- it's more than likely mathematical rounding.

Simply put, the current wind product is a useful tool, but has its limitations and is rather easy to misinterpret. The NHC is working on a new tool -- http://www.nhc.noaa.gov/text/refresh/MIAPWSAT3+shtml/DDHHMM.shtml as a current example for Rita -- to try to better represent a storm's potential impacts, with one of the simplest improvements being that the track forecast and wind probabilities now both go to 5 days apiece. In this product, as of 11p Monday, Galveston, TX has the highest probability at 44% for cities in terms of the second landfall. But, you see some non-zero probabilities for other areas, just in case the storm does something unexpected.

Look at it this way: I don't believe the storm has a shot to impact Tallahassee, where I am located. But, say the unexpected were to occur and it turned northward. There is precedent for something like that, it is within the realm of accumulated track error (not for one forecast, but a series of them), and thus you see a non-zero porbability (here, 7%). A similar interpretation can be applied to the older product, just keeping in mind the limitations inherent to any such product.