The different hops utilization models such as Tinseth, Rager, Garetz, etc., all use a different mathmatical algorithm to reach their estimate of the IBUs you'll get from X hops with Y AA% boiled for Z time. There is a podcast somewhere online (basic brewing radio?) where they entered six different recipes and used various hops model for each recipe. They got wildly different IBU forecasts (one from each model), and then sent the actual beer to a lab and measured the real IBUs achieved.
The results were humorous. No model was clearly the best. Some were better estimating IBUs with lots of late additions, other were better with low-gravity recipes, etc. Some estimates were off by ~40 IBUs!
Two things that may help: ProMash defaults to Rager, BeerSmith defaults to Tinseth, so if you're using a recipe that someone wrote in ProMash it might be Rager's model. Second, book and magazine recipes rarely state which hops model they used, so try switching models in BeerSmith and see if any of the main three models gets much closer to the printed recipe. (Tools\Options\Bitterness)
I actually asked Jamil about this a while ago and he recommend picking one model and sticking to it. It's all relative, so if you use Rager and 45 IBUs is not bitter enough, increase the hops until you find the taste you seek.