I tried to do "true" literate programming but found the tools inadequate. Specifically, for some snippets of code I wanted to show them in more than one place in the document but that lead to all kinds of trouble. I found other things cumbersome as well. Some of that may be my inexperience, or failures of the
specific tool set I chose.
Currently I use a bastardized form of literate programming that would not qualify as "real" literate programming. I put the code in the order the code belongs and I embed the document within comments in the code. This leaves the source executable for scripting languages, or directly compilable for compiled languages. We made a tool to flip the source "inside out" to produce a document (TeX/PDF,HTML).
According to LP people I am doing it wrong and not really getting the benefits. That's pretty harsh, but I've managed to live with myself somehow. What we end up with reads like a document, looks great as a web page or printed, and provides encouragement to write meaningful text descriptions. The biggest part to stress is the very last bit: writing meaningful text. Since you produce a document you expect it to read like one. This discourages stupid comments stating the obvious and encourages you to talk about reasons why and other meaningful things.
Please keep in mind that this is just my opinion, coming at LP without the benefit of working with someone who already knew it well.