I can't really relate to all the other great IDE's out there for Python, but in the scientific context I think Spyder makes a lot of sense. It tries to create a comfortable environment for scientists with a limited background on computer science. Especially for the more GUI oriented users, it is great to have a visual on the variables in the workspace with the variable inspector (like Matlab has). Debugging toolbar, object inspector showing rich text documentation (including MathJax rendered formulas), running a multitude of Python or IPython consoles, syntax checking etc makes it for my line of work as a scientist a very decent environment.

There was a nice tutorial given on the SciPy 2013 conference [0] by Carlos Cordoba (the main developer these days for Spyder I believe). What I also think is exciting is that he shows some ideas about how to integrate the IPython notebook in Spyder (last minutes of the talk). Would be interesting to see when that feature further materialize. I think that is a potential killer feature. There are more interesting idea's on further improving Spyder, please ping the mailing list if you are interested in helping out :-)

Besides python(x,y), the WinPython distribution [1] can sometimes offer some installation releave for Windows systems (don't have experience with it as a Linux user though).

@Rickasaurus: When running Spyder straight from source, you can have multiple instances running. Not sure in what context that would be useful besides developing Spyder with Spyder [2] though: python bootstrap.py -- --new-instance

@Derbasti: Spyder does work with Python 3 [3] for the older 2.1 branch. For 2.3 I believe full Python 3 support is also eminent.

[0] http://pyvideo.org/video/2113/the-advantages-of-a-scientific...

[1] https://code.google.com/p/winpython/

[2] http://code.google.com/p/spyderlib/wiki/NoteForContributors

[3] http://spyder-ide.blogspot.dk/2013/01/spyder-v2114-supports-...

Disclaimer: I am a tiny and modest Spyder contributor.

Edit: fix layout