Uncle Bob writes about why rewriting an existing Big Ball of Mud is probably the worst thing you can do:
http://blog.objectmentor.com/articles/2009/01/09/the-big-redesign-in-the-sky
It’s depressing to think that many of the projects I have been brought in on over the past few years were grand rewrites. Technically they have been relatively successful, but probably viewed by the business as time-wasting and expensive. It’s time for this madness to end! Pick up a copy of Michael Feather’s book and start scrubbing!
Tim Ross - Software Developer 
I’m really starting to think that a lot of this is down to peoples inability to work with legacy code and understanding of how to refactor a mess into something maintainable.
@Mike, I think this has a lot to do with being unsure about how to change legacy code without breaking it. Many devs probably think it’s easier just to start again, which is almost never the case! I think there is a lack of knowledge on how to break dependencies and effectively test legacy code, in order to safely refactor and introduce changes.