Finally I think I (kind of) get the idea on how to do DOF in Blender.
1. Allow LIMITS of your camera to see the start, end and yellow focal point crosshair.
2. Adjust Start / End and more importantly the DofDist
3. In NODES window, Hit "Composite Nodes" (face icon) and "Use Nodes"
4. Add "Defocus" node.
5. Take off "Preview" and "No Z Buffer" from this node because they tend to stuff up workflow times.
6. A new "Fstop" should show up. Take it down to about 16
7. In rendering, under the "Anim" button, hit "Do Composite"
Now when you render (hit F12) you should get the DOF you wanted.
Has Blender "moved forward" from this approach? It seems like an awful lot of steps for what should be a simple camera setting. Most other packages I've used have something like steps 1, 2 and then 7... no nodes or anything needed.
...also: if I decide my DOF is wrong (want more blur, or less at a particular point), is there a way I can rerender the settings only without restarting a full render pass for the initial image?