I think there were a few factors, the combination of which was greater than the sum of their individual weights.
One is simply timing: Django appeared right as the first big wave of Rails hype was ramping up, and so it was immediately portrayed as being sort of "Python's answer to Rails". That resulted in a not-insignificant number of eyeballs on the project almost from the start. The fact that Adrian was at the "Snakes and Rubies" meetup in Chicago and got to take part in side-by-side talks about Rails and Django did a lot for that.
Another factor is that Django is and always has been a single-package install (well, not quite: you still need a database adapter, unless you're on Python 2.5+ and using SQLite, but close enough). The non-Zope alternatives, which all focused on leaving component choices in the developer's hands, required quite a bit more work just to get to the point where you could do a basic tutorial: you'd need to go hunting down an ORM, a template language, etc., etc. and get them all installed and configured. Though that's gotten much better over the years, I think the lingering memory of that still has an effect.
And Django came out of the gate with documentation that (if I may say so myself) was far above the usual standard for open-source projects, and has only gotten better over time. The tutorial, for all its many faults, hits on a number of the high points that make Django useful, and the remainder of the documentation has always been of good quality, mixing both API reference and important "how to" bits as needed. This produces a good out-of-the-box experience and helps with the post-tutorial learning curve (something which has always plagued Zope).
I also think there's a perception -- rightly or wrongly -- that, say, Pylons or Werkzeug really are better for experienced developers who already know their way around WSGI and the Python web ecosystem; the fact that they tend to be strong choices for taking your existing favorite libraries and plugging them together is the source of this, I think, and perhaps nudges some newer folks over toward Django's integrated approach. The flip side, of course, is that a lot of people who'd be better off learning more up-front before trying Django don't do that ;)
Finally, I think there's something to be said for the way Django's been marketed, which is to say that it really wasn't marketed for a long time, or at least not in the sense that, say, Rails was marketed. Until Django 1.0 landed, the "marketing" effort mostly consisted of people blogging (and there were some notable incidents where people were asked to tone it down a bit), talks at PyCon and then mostly just improving the framework, building cool things with it and letting the results speak for themselves. Now, of course, in the post-1.0 world we have the DSF and DjangoCon and business-oriented consultants doing training sessions and lots of books and all the rest, but that's all still quite new.
I expect that there will be a backlash, just as there's been with Rails, and in fact I think it's been brewing for a while and has already started. But up until now I think the factors I've listed here are at least the major ones behind the consistent, steady growth in popularity Django has seen since its initial release.