Visits is the one to look at.
Hits and file requests can be a minefield to newcomers to web stats.
An example for you:
Say your index page has a banner, a picture in the main content window and perhaps a small logo at the bottom of the page in the footer area next to your copyright statement. It is also formatted using a separate CSS [Cascading Style Sheet] document and has a repeating background image tile.
When I visit your site and view your index page I've not only "requested" the index.htm document, but I also needed [and got] the CSS document, your banner image file, the logo image file, the repeating background image tile and the picture in the main content window. One page: many files. That's how it works.
If you have different images on certain pages, or rollover images for your buttons, or any one of a number of other things going on in there, the file requests go through the roof.
Imagine a set of rollover navigation buttons. Say you have six buttons [home, our services, about us, contact us, find us, our portfolio] and you have a rollover effect going on. Let's say you have a different image for each button in its various states [static, hover, onclick and even visited] suddenly your six buttons can need a lot of different files to support them and so your file requests shoot up in your web stats since each individual file has to be included with the document being viewed.
Look for unique visitors rather that hits and file requests.