Since the late 1970's, redshift surveys have been vital for progress in understanding large-scale structure in the Universe. The original CfA redshift survey collected spectra of 20-30 galaxies per clear night on a 1.5 meter telescope; over a two year period the project added ~2000 new redshifts to the literature. Subsequent low-z redshift surveys have been up to an order of magnitude larger, and ongoing surveys will yield a similar improvement over the generation preceding them. Full sky redshift surveys have a special role to play as predictors of cosmological flows, and deep pencil beam surveys have provided fundamental constraints on the evolution of properties of galaxies. With the 2DF redshift survey and the SDSS survey, our knowledge of the statistical clustering of low-redshift galaxies will achieve unprecedented precision. Measurements of clustering in the distant Universe are more limited at present, but will become much better in this decade as the VLT/VIRMOS and Keck/DEIMOS projects produce results. As in so many other fields, progress in large scale structure studies, both observational and theoretical, has been made possible by improvements in technologies, especially computing. This review briefly highlights twenty years of progress in this evolving discipline and describes a few novel cosmological tests that will be attempted with the Keck/DEIMOS survey.

To appear in the proceedings of the ESO Workshop on "Mining the Sky", Munich, August 2000, ESO Astrophysics Symposia, in press