Frank Weinhold, the Godfather of natural orbital theory, has a description of natural orbitals at http://www.chem.wisc.edu/~nbo5/web_nbo.htm

The key to the NBOs is that they are few-atom-centered rather than the many centered MOs you would generally find for a polyatomic - they look much more Lewis-like or (shudder) hybrid-like.  But they are not classical hybrid orbitals.  For example, in methane, the NBOs come out looking similar to hybrids, but in water, the "lone pairs" remain an oxygen p orbital and, largely, the O 2s orbital, while the O-H bonds are degenerate and 2c-2e.

There are certainly many papers that use NBO theory as it reveals a number of interesting chemical concepts that are difficult, if not impossible, to tease out by conventional MO theory.  For example, there is strong evidence from Weinhold's work that the rotational barrier in ethane arises not from steric effects (the Hs are fairly far from one another) but from a hyperconjugative interaction between the CH sigma bonding orbitals on one C and the CH sigma antibonding orbitals on the other C - an interaction that is optimal in the staggered conformation.

Thanks for the info, Flick, and thanks for contributing to VIPEr! Organic chemists tend to like NBOs because they are more hybrid orbital-like. But you did a nice job of pointing out the differences.

The web link you provided shows interesting examples with diborane and ethane. From an inorganic chemistry perspective, do you know what happens when this approach is used, for example, on octahedral transition metal complexes?