Wow, these have been some really interesting answers and approaches. 
Great stuff.

> Glen said:
> Excellent point!  Here's my revised version that should not have this problem.

*snips maze output*

I don't know if this is just coincidence, or that I might being seeing
something that really isn't there ... but looking at your output in
the proper fixed-width font makes me wonder if your algorithm is truly
random.  I can't help but look at that maze and see a very heavy
presence of diagonal paths ... ie. a high occurance of being able to
travel right, down, right, down, right, down, right, down, etc.  Or is
it just me?

I took a look at your source, and nothing jumped out at me as to why
this might be the case...

Matt