I must admit that I've inherited two messes where someone used NHibernate as a replacement for SQL. In both cases, the schema was excruciatingly simple but the code performed excruciatingly poorly.

In once case the project was canceled, in part because it was so late due to the developers not knowing how to use a database. In the other case I put my foot down and removed NHibernate. The schema was so simple that it was just easier to put a few extra minutes into boilerplate code than to put time into learning a new thing.

I'd really like to see a good writeup about the use cases that tools like Hibernate excel at. The problem is that, in both cases, I had to work with a high-level manager who had very bad assumptions about what Hibernate can and can't do.