Thermoelectric devices offer the promise of converting heat to electricity, by means of the thermoelectric effect, in which a temperature gradient applied to a material produces a voltage gradient, which can be used to drive current. The efficiency of a thermoelectric device is quantified largely by a single parameter, the dimensionless "Figure-of-Merit" ZT. Currently available commercial thermoelectric devices, based on Bi2Te3, have a maximum ZT of roughly unity and are used mainly in niche ambient temperature applications such as beverage coolers. However, intensive research over the past decade has succeeded in producing materials with ZT values approaching 2 at high temperature, which makes them candidates for widespread applications. In this work I will discuss several recent ORNL findings in this area: - the discovery (PRL110, 146601 (2013)) that cubic materials, including such high-performance thermoelectrics as PbTe, PbSe and SnTe, can be made to behave as if they are low-dimensional from the electronic point-of-view, offering a real-world realization of the low-dimensionality enhanced performance envisioned by Hicks and Dresselhaus.- the finding (PRB87, 045317 (2013)) that the origin of the effectiveness of "bulk" nanostructuring in reducing thermal conductivity (a key factor for useful thermoelectrics) lies in acoustic impedance mismatch scattering as a result of sound speed anisotropy, which can be directly quantified from the measured elastic constants; and- the result (PRB87, 045205 (2013) that the mineral costibite (chemical formula CoSbS), comprised Entirely of inexpensive abundant elements has the potential for high thermoelectric performance (ZT >1) in an elevated temperature range around 800 K, and that a related mineral gudmundite(FeSbS) may also have similar potential as a high-temperature thermoelectric. I close the talk with a discussion of the likely applications of thermoelectrics in the near future and where I see the field heading.