The Cryogenic Dark Matter Search (CDMS) attempts to directly detect galactic dark matter via keV-scale nuclear recoils in semiconductor detectors located deep underground. I will present recent results from the current phase of the experiment, SuperCDMS Soudan, in which an array of Ge detectors has been used to search for low-mass dark matter, such as is predicted by models relating the baryon asymmetry to the dark-matter abundance. Each cylindrical crystal has interleaved phonon and ionization sensors on both flat faces, enabling improved background-rejection capability relative to our previous detector design (CDMS II). We exclude new parameter space for low dark-matter masses and strongly disfavor a dark-matter interpretation of the excess reported by CoGeNT, which also uses a Ge target. To extend to lower masses and cross sections, we will build the SuperCDMS SNOLAB experiment, recently selected as part of the next generation (G2) direct-detection program. Operating more and larger detectors with lower energy thresholds in a lower-activity shield promises greatly increased sensitivity (~3 orders of magnitude at 5 GeV). In particular, I will discuss methods of radon mitigation and assay that may be used to prevent radon-progeny surface contaminants from dominating the expected backgrounds.
