Okay, I developed an isoctatic separation on a traditional 150mm x 4.6mm i.d. column, looked good. The manufacturer made up three 150mm x 3mm i.d. columns for validation studies; the flow rate was adjusted to about half to keep linear velocity/retention times the same. The resolution was actually worse on the 3mm i.d. columns. Halving the injection volume from 10ul to 5ul didn't help. The buffered mobile phase was unchanged, same reservoir. Have any others observed this or can explain this?

Your loss of resolution probably comes from extra-column volume which is trivial compared to the volume of a 4.6mm column but highly significant compared to the volume of your 3mm column. Minimize this and your resolution will improve. You need to eliminate at least 60% of your extra column volume. You may need to reduce the size of your flowcell as well.

I fully agree with the previous statement. When scaling down to narrow diameter columns, extracolumn band spreading must be minimized to maintain the chromatographic resolution. For isocratic separations, the bandspread must be minimal before and after the column. For gradient separations, the bandspread must be minimal throughout the system and is even more important after the column. So, to maintain your desired resolution you will have to make some changes to your system. Some things you can include changing tubing inside diameters to 5/1000 i.d., making these tubes as short as practical, removing any unions that are in the way, lower volume flow cell for the detector.

Just out of curiousity, why did you switch from a 4.6 to a 3.0? Of course I agree with the above as well-you should always keep the dead volumn to a minimum. I have found that typical you do not need to replumb the LC sytem when going from a 4.6mm to a 3.0mm assuming you had it optimized to begin with.
If you were looking to increase resolution, the column parameters to change would be the particle size (going to a smaller particle) or length (going to a longer column if it is available).
Another thing you may consider is what the effiecencies are on the two columns. If the manufacturer is not use to packing that particular column id you may get a column that barely passes the QC specs whereas the original 4.6 id column may always pass on the high end of the specs. Having worked for a column manufacturer in the past I found this to be true for certain materials-they packed really well in the 4.6 size but had a high failure rate in the 3.0 id for some reason (we had to discontinue offering certain materials in the 3.0 size since it was nopt worth all the trouble). I would suggest you contact the column supplier and seek out more information on the columns if they did not come with a column spec sheet. I would expect a small decrease in resolution but without seeing the chromatograms it is difficult to judge if what you see is normal.
Good luck!