Google, who's been determinedly digitising the world's books, thought it might be prudent to figure out just how many books are actually out there to begin with. They made a special algorithm, natch, and came up with 129,864,880.
This post on Inside Google Books blog explains the whole process. First, you have to decide what exactly counts as a book, and Google's notion is pretty similar to the set of texts that ISBNs - International Standard Book Numbers - are attributed to.
But ISBN is a relatively new standard, coming around in the 1960s, and a relatively Western one, so there's a huge selection of books - old books and foreign books, largely - that don't have ISBNs.
So Google leapfrogged ISBN and pulled down metadata from some 150 sources, including libraries, WorldCat, national union catalogues and commercial providers, and then eliminated the duplicates. That got them to 600 million records. After eliminating more duplicate holdings, microforms, maps, audio recordings and other assorted non-books, and after adjusting the number to account for serial publications, the number is 129,864,880. Hey, I've read a few of those! [Inside Google Books via CrunchGear]
Image credit Patrick Gage