PyMol is a python-based molecular viewer which is able to generate beautiful pictures with excellent light effects. It is thus good for making publication-quality images. The easiest way to get hold of this program for free, is to download the 0.99 version from sourceforge page. This is an old legacy version, but for most applications you should find this perfectly adequate. A wiki manual is available.
MolMol was written specifically with NMR users in mind and is therefore good at dealing with ensembles. It is also able to display certain types of NMR restraints and violations and has many additional functions such as being able to calculate mean structures, RMSDs, show Ramachandran plots etc. Loading non-protein/peptide molecules can unfortunately cause problems and the manual can sometimes seem rather cryptic which makes it hard to get the most out of the programme unless you have an advanced user to hand who can help you. MolMol can be operated via a GUI, but is also possible to write scripts to carry out a series of commands.
Rasmol is quite a nice programme for looking at molecules quickly. Unfortunatley, the normal version is not able to deal with ensembles very well (or possibly at all). It is also not brilliant for generating high quality graphics for publication. However, it is developed quite a bit in the mean time and the newer versions might be more useful to NMR users.
I know very little about this, but everyone who uses it raves about it.
This is a version of VMD which the NIH NMR groups have adapted to work with XPLOR-NIH. It is thus able to show restraints and violations and other such things which are useful for NMR spectroscopists.
This is quite a nice programme for looking at proteins. It has various exciting functions such as being able to mutate a sidechain, change torsion angles, add hydrogens, very easily measure inter-atomic distances and a magic-fit which will fit non-identical proteins. But it has no NMR-specific functions.
This is a project which is part of CCP4. It is fairly intuitive to use and has good online tutorials. It can deal with NMR ensembles, with ligands and with crystal lattices and is thus fairly versatile. But as it is written by crystallographers it does not have any facilities for looking at NMR restraints and suchlike.
This is actually a programme for modelling crystal structures, but it has quite nice graphics. It doesn’t handle hydrogen atoms very well, so it isn’t ideal for NMR structures. However, it is possible to look at crystal strutures with the symmetry molecules included – excellent for investigating crystal contact areas!