In Python 2.7, there are many patterns of the form

for x in mapping.(keys()|values()|items()):
    #code
when the iterator version of those method could be used.
Is there any reason for using those methods?

I see that in some other parts of the code the iterator version is used. Here is the summary:

       Non iter version Iter version
Keys   160              21 
Values 35               23
Items  249              79

I sued the following command

$ egrep -R "for.*[.]iterkeys\(\)" . | wc -l 


What's the position about this? Does it worth it to fix this?

Thanks for you interest in improving Python. As noted in the response to one of your other issues, we generally do not do style cleanups just for the sake of style cleanups.  In addition, our policy for current maintenance branches, like Python 2.7, is to apply bug fixes only.  As there will be no further feature releases for Python 2, anything other than a bug fix would only be a candidate for the next Python 3 release (Python 3.3).

See the Python Developer's Guide for more information:
http://docs.python.org/devguide/devcycle.html#maintenance-branches

skreft: since you were asking for a reason why the code looks the way it looks - the code in many cases predates the introduction of iterkeys and friends.

I personally find the names iterkeys/itervalues/iteritems fairly ugly, and rather avoid them for prettiness reasons, unless there is a real problem that requires using iterators. Thankfully, the issue was resolved in Python 3, where .keys() does the right thing (i.e. returning a view).