I'm late to the party on this one, but got it right from reading the Inside SQL Server 2005 books. Kalen Delaney stressed that it was balanced. A couple of comments.

1. "Balanced" is significant because it means the structure of the indexes tries to keep accesses to different "leaves" as close to one another as possible. Some people give warnings about what particular data loads might do to a b-tree structure, but they aren't talking about "balanced" indexes.

2. People who say that it is unknown should have the humility to say that it is unknown to them. Obviously if it is being published in official Microsoft publications, it's not being hidden.

3. If things like this aren't in BOL, they ought to be.