Can't believe I've got to the end of my PhD before asking this one! I want to cite a work whose copyright date is 1996 but whose first paperback edition (the one I'm referencing) is 1999. Which date is correct?
I have actually asked a librarian about this issue as it came up for me in the past.
She told me to cite the exact copy that I am reading. If what you are reading is the first paperback edition (usually the inside left initial pages will have information about this), then do cite that.
The purpose of giving details in the citation (rather than just saying "according to Joe Bloggs") is to allow other people to find the cited work. Year of publication is something that could plausibly help somebody find the work; copyright date is not, so cite the year of publication.
You should give the bibliographic information that the book gives you. Nothing more, nothing less. You should definitely cite the year that the publication itself puts as the year of publication (usually after a copyright symbol). Here I clearly disagree with the other answers given here. Here are a few reasons for this advice.
One, this will be the year that most, if not all, library databases will put down as the year of publication. You will confuse a lot of people if you put
1999 in your reference list when there is no
1999 entry for that book in the library catalogs.
Two, it is commonplace that the "year of publication" is different from the year it actually appeared. Sometimes journal articles from 2004 will often not appear until 2005, other times journals and publishers will deliberately back-date the publication by one year. I have books where the publication year is
n, but everybody knows it actually appeared in
n+1. It's not your job to be a bibliography detective and tell people when it really appeared. Your references should just say what the publication says.
Three, the reasons most paperback issues don't change the original year of publication is that the paperback copy is identical to the original hardbound copy. They are usually just photographic reprints. In many cases they won't even inform you when the reprint occurred - it'll just say something like "3rd reprint" or "3rd print". Again, you shouldn't be a detective and try to figure out when that reprint appeared, since it is of no relevance to the readers if the reprint is identical to the original.
So this is what you should do:
If the reprint/paperback issue gives the original year of publication, but also tells you something about it being a reprint, then add the information about the reprint as a note, e.g.
Johnson, John. 1996. A book about nothing. Oxford University Press. 3rd reprint, 1999.
If you really care (like I do), get the original. I know most people don't bother, but I always do :)