Google's Matt Cutts recently posted a video detailing how much control Webmasters actually have over the info that appears in their search snippets, and Jason Lee Miller outlines the highlights with this post.
The first piece of info is the title. Using Cutts' example of Starbucks, Miller points out that the coffee chain could have chosen to use "Starbucks Coffee" as its title instead of "Starbucks Homepage" -- but that it was still clearly their choice.
For the meat of the snippet itself, Google often pulls the text directly from a site's meta description tags, and if no tags are relevant or available, will grab info from directories like dmoz. You can keep Google's spider from indexing specific content for the snippet by adding a "nosnippet" tag to the HTML segment.
The video also includes tips for how to tell Google what pages to cache (or not), but Cutts makes it clear that site owners do have less control over what content gets listed as a sitelink.