Here is the script content for reference:

mkdir c:\work
cd c:\work
mkdir rootrepo
cd rootrepo
hg init
echo test > Test.txt
hg add
hg commit -m "init"
cd ..
hg clone c:\work\rootrepo clone1
hg clone c:\work\rootrepo clone2

cd clone1
echo test2 > Test.txt
hg commit -m "test2"
hg rename Test.txt test2.txt
hg rename test2.txt test.txt
hg commit -m "rename"
hg push
cd ..

cd clone2
echo test3 > Test.txt
hg pull
hg update

I was able to reproduce this with a smaller script:

hg init test
cd test
echo test > Test.txt
hg add
hg commit -m "init"

hg rename Test.txt test2.txt
hg rename test2.txt test.txt
hg commit -m "rename"
hg upd 0

echo test3 > Test.txt
hg upd

The output of the second 'hg upd' is:
merging Test.txt and test.txt to test.txt
0 files updated, 1 files merged, 0 files removed, 0 files unresolved

But the file is deleted.

Fixed by http://selenic.com/repo/hg/rev/5b3383ea67d2
Matt Mackall <mpm@selenic.com>
merge: avoid unlinking destination of merge when case changes (issue2715)

(please test the fix)

--- Bug imported by bugzilla@serpentine.com 2012-05-12 09:18 EDT  ---

This bug was previously known as _bug_ 2715 at http://mercurial.selenic.com/bts/issue2715
Imported an attachment (id=1536)
Imported an attachment (id=1537)