While there is convincing evidence that the central regions ($r \ll R_e$) of early-type galaxies are dominated by stars and that the outer regions ($r \gg R_e $) are dominated by dark matter, the structure of early-type galaxies in the transition region (a few effective radii $R_e$) between the stars and the dark matter is unclear both locally and in gravitational lenses. Understanding the structure of galaxies in this transition region is a prerequisite for understanding dark matter halos and how they relate to the luminous galaxy. Potentially the best probe of this region is the sample of $\sim 80$ strong gravitational lenses. I review the determination of mass distributions using gravitational lenses using image positions, statistics, stellar dynamics, time delays and microlensing. While the present situation is confusing, there is little doubt that the existing problems can be resolved by further observations.