Since learning about them some nine years ago, I’ve found User Stories to be much handier for expressing what the system should do than the monolithic requirements documents I dealt with in the past.  User Stories have a number of advantages:

• They can easily be shuffled into a different order.
• You generally know (by the acceptance criteria) whether or not they’re done.
• They don’t contain a lot of duplicated or contradictory detail.
• The detail gets elaborated when it’s needed (rather than a long time prior, letting the details get out-of-date and/or forgotten).

I’ve created a two-page handout on writing and splitting User Stories.  I’m publishing it here in case it helps others.  And I would welcome any feedback on it.