Sounds like you're running into servers that like to 
intermittently drop connections.  I had a similar problem 
when I reaped <http://members.aol.com/bblum6>.  Sometimes 
HTTrack would be able to grab the whole html page, 
sometimes not...it seemed random when it would succeed.  I 
observed the same behavior in my web browser, IE 6.0, and 
figured the problem was with the web server because trying 
to reload the page a few times tended to get the desired 
results.

I guess the thing that remains to be done is tweaking 
HTTrack's update mechanism not to fall victim to flaky 
webservers like that membes.aol.com server.  (By this I 
mean that when I'd use the HTTrack update website option, 
I'd sometimes end up being counter-productive as HTTrack 
would detect 'changes' in some web pages that were actually 
not changes but errors in the download of the html).  Is 
there a way to detect the broken/terminated connections?  
Detecting them might lead to better logic in the HTTrack 
engine about whether or not to replace a previously 
downloaded copy of a page.

Oh, I didn't try the HTTrack file-by-file confirmation 
(Q/A) option in this case because the site wasn't all that 
big and the broken files were few.