Understanding Hash Join performance

A query I has spends a long time on Hash Joins (and Hash Left Joins).
I have a few questions:

1. When does Postgres decide to do a Hash Join, over another type of Join?
2. Do Hash Joins normally perform poorly?  What can I do to speed them up?
3. What can I do to enable Postgres to use a faster type of join?

If there's a good resource for me to read on this, please let me know.

Thanks!