The determination of chemical abundances from stellar spectra is considered a mature field of astrophysics. Digital spectra of stars are recorded and processed with standard techniques, much like samples in the biological sciences. Nevertheless, uncertainties typically exceed 20% and are dominated by systematic errors. The first part of this paper addresses what is being done to reduce measurement errors, and what is not being done, but should be. The second part focuses on some of the most exciting applications of stellar spectroscopy in the arenas of galactic structure and evolution, the origin of the chemical elements, and cosmology.