First: I am the author of PyUseCase, and I was no longer playing with Dogtail ...
In different ways, the tools are different.
Dogtail works through the accessibility interface in Gnome on Linux, and PyUseCase works with GUI tools (PyGTK, Tkinter, SWT / Eclipse in the current version, plus Swing from the upcoming new version)
PyUseCase is very difficult to use for non-programmers. User interface actions are defined in the domain user language, statements are replaced by the creation and comparison of simple text descriptions. It also contains a recorder.
Dogtail is the more traditional paradigm of "writing Python code, calling APIs, saying things about what you get" is probably the more familiar way of doing things if you are used to programming and unit testing.
/ li>PyUseCase tries to easily change tests in bulk when changing the GUI. The verification code is super repairable because it is not. You get instead a “UI map file,” which is just a definition.
Finally, I'm not sure how active Dogtail is. The last time I looked, it seems that the last completion was in 2009, but the looks can fool ... If you want something like Dogtail, I would suggest looking at the Linux Desktop Testing Project (LDTP), which is pretty similar on the concept, but it seems more active.
PyUseCase is active anyway, so the two of us work on it full time. It works great on our real graphical interfaces, but its maturity varies between different tools.