How about using a biotinylated oligo for the counterstrand, so that you could fish it out with as streptavidin column after denaturing the PCR product ?

I would go for the biotin labeling as well.

This could be useful:

http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0094752

thanks for your answers. Introducing a biotin is an additional expense that I am specifically trying to avoid - I'd love to go all DynaBeads and forget about it!

In my case, I figured that the dsDNA product can participate in the addition reaction via the 5'-NH2, and then the overall product can be treated with 5'->3' Exonuclease, as the strand of intererest will have its 5' inaccessible.

Exonuclease should work but then you might need to get rid of the enzyme, depending on your application. The biotin-labelled oligo should be less than 90 USD, exo enzyme might be about the same.

I totally agree with the biotin answers but if cost is the main factor perhaps you could run a pcr (assymetric) with excess forward primer and limiting reverse primer then run lots of cycles when only the primer producing your strand works. there will remain some reverse strand but hopefully in amounts small enough to not interfere with your application

yes, I thought of the asymmetric PCR, but that would then involve gel band purification to extract the ssDNA. Seemingly a more robust technique, albeit quite time-consuming. Thanks again for your contributions