The cold dark matter scenario predicts that a large number of dark subhalos should be located within the halo of each Milky-way sized galaxy. One tell-tale signature of such dark subhalos could be additional milliarcsecond-scale image splitting of qu
asars previously known to be multiply-imaged on arcsecond scales. Here, we estimate the image separations for the subhalo density profiles favoured by recent N-body simulations, and compare these to the angular resolution of both existing and upcoming observational facilities. We find, that the image separations produced are very sensitive to the exact subhalo density profile assumed, but in all cases considerably smaller than previous estimates based on the premise that subhalos can be approximated by singular isothermal spheres. Only the most optimistic subhalo models produce image separations that would be detectable with current technology, and many models produce image separations that will remain unresolved with all telescopes expected to become available in the foreseeable future. Detections of dark subhalos through image-splitting effects will therefore be far more challenging than currently believed, albeit not necessarily impossible.
