From the help files:
4.Set Orientation to one of the following:
◦Prefer Touch — Presents a touch constraint when touch and align solutions are both possible. (Touch constraints are more common than align constraints in most models.)
Prefer Touch presents an align constraint when a touch constraint would over-constrain the assembly.
◦Touch — Constrains objects so their surface normals are in opposite directions.
◦Align — Constrains objects so their surface normals are in the same direction.
◦Infer Center/Axis — Specifies that when you select a cylindrical or conical face, NX uses the face's center or axis instead of the face itself for the constraint.
Thanks. That helps with the difference between Touch and Align as they relate to surfaces. The reason for the question is because I come from an I-deas background. If I had two plates with a hole in each, I would use a "coincident-collinear" constraint applied to each centerline of the two holes to "align" them, and then would use the same constraint again, but this time would select the surface on one plate and the end of the centerline of the hole on the other plate to make them coincident. In NX, I don't think you can select the end of a centerline (or centerpoint of a hole) for your constraint. So in NX, I use "Align" on the hole centerlines, and then "Touch" on the two surfaces of the plates, but that should result in overconstraint. So, I guess from now on, I should use "Prefer Touch" for the second constraint.
Let me know if that is wrong.
I use prefer touch for 90% of my constraints. face to face, and hole centerline to centerline.