Hi Larry, sorry I didn't reply until now. Somehow I didn't notice your comment before.
There are a few variables in predicting ISS passes. For instance, when is the sun considered low enough below your horizon for you to be able to see the ISS? And is the start of a pass timed from the moment it comes up from behind the horizon, or when it is at the minimum altitude (in our case, 11 degrees)?
A difference of two minutes is unusually big though, I think. If you see that again, let me know. It would be interesting to compare results in these cases. I have benchmarked our data to that of Heavens-Above.com and those matched really well.
It could be the height. Twisst does not take the elevation of your location into account at the moment. I think our system uses a default of 20 meters. That could account for some difference.
Maybe the locations you enter at the different sites accounts for some difference too. Twisst geocodes the location in your Twitter profile. If that process results in slightly different coordinates than you use at the other sources, than you would expect a couple of seconds difference here and there.