I just came across another post where OP was asking if author's name should even be in the file header and seems that at least 2/3rd of people who responded said that the name shouldn't even be listed and that you should use version control to simply keep track of who changed the file. Don't know what happened to that post, but now I can't find it. <-- (hence anonymous "OP")
Personally, I find author listed in the file header to be useful but for a slightly different reason (and this may not relate to others in their environments). Even though we try to practice community ownership and often work on various parts of the project, we tend to have few team members who know certain areas of the code much more intimately than others. So when someone (especially numerous contractors that come and go) open a file that they've never seen before, author becomes the go-to person. He may not be the only contributor, or even majority contributor, but having his name at the top, he acknowledges to have certain responsibility in distributing knowledge/information about the code to the rest of the team. We can list more than one person in the header is multiple people have indeed contributed and feel responsible.
I do find it frustrating when I have a question about a file and have to resort to version control to identify primary, or most knowledgeable person. Then end up going from one guy to the next as they all deny really knowing what the code does... they just had to go in and fix a bug or two.
This practice works in our team because we don't have hand-offs. Unless a person quits, or moves to a different team, that code/project will stay with the person and with our team. Obviously if people who maintain the code are not the same as those that write it, then nobody would care who was listed in the header.
So in light of my view on file headers, I'd say if you changed 80% of the file and you feel like now you are the go-to guy for any questions (and you probably should feel that way), yes, go ahead and update the file header to have your name on it. If you feel bad about removing previous person, you could leave their name there as well, at least for the time being. You can always ask original author and I'm sure they won't mind one bit that you changed the name, since I'm assuming there's no hard feeling about you changing 80% of the file itself.
UPDATE: Found that post. Have no idea how I managed to pull something back from August. I just finished reading The Pragmatic Programmer and in the last chapter authors talk about signing work and accountability (the other post mentioned it, that's why I looked it up). The book makes perfect sense and now that I think about it, maybe we should introduce team policy that whoever is listed as the author, should be included in all code reviews of the file in question. Doesn't matter who changed the file last or most in SVN, the author is the owner and is the keeper.