Lots.
There's really not a lot of costs that can add up for reuse of these stages. It'll require a bit of cleaning up, a few new bits and pieces replaced and added, and a lot of inspection work. But most of the cost of the stage is in manufacturing the engines and tanks, so it should add up to enormous savings, even if it's relatively costly to reuse each stage compared to the theoretical limits.
Edit: the flip-side is that the reliability and robustness improvements from reusability may be as big a win as cost. Currently it costs tens of millions of dollars to launch a rocket to orbit, which means it's almost never done except as part of a paid launch. Moreover, despite the seemingly high number of launches very few of those launches represent expanding the test-envelope much, every single launch is typically straight down the middle of the performance envelope, to maximize the chance of success. That results in learning very little about these vehicles despite how much they've been flown. By introducing reuse and dropping the cost of flight it may become possible to do real test programs, which would make it possible to determine the flight envelope characteristics of vehicles and help lead to improving designs over time.