The history, philosophy, and application of the concept of the Global Stratotype Section and Point (GSSP) are reviewed. Geochronologic units defined by GSSPs serve as practical classificatory pigeonholes for the subdivision of geologic time. Accordingly, the main factor involved in the definition of GSSPs must be global correlatability. Early opposition to the GSSP concept centered around the desire for a traditional biochronologic time scale defined conceptually in terms of palaeobiological events, but such time scales are inherently unstable and thus unsuitable for the use of all geoscientists. The GSSP concept is also generally incompatible with the desire for ‘natural’ geochronologic boundaries. GSSPs have been defined mainly on the basis of biostratigraphic guiding criteria, but magnetic polarity reversals and chemostratigraphic and cyclostratigraphic horizons are now playing an important role. Most primary guiding criteria used to place a ‘golden spike’ will eventually become problematical in some way, so GSSPs should be defined so as to be correlatable by as many different lines of age-significant information as possible. The ‘Global Standard Stratigraphic Age’ (better renamed ‘Standard Global Numerical Age’) is a numerical analogue of the golden spike. Numerical definitions are currently appropriate for the formal subdivision of the Precambrian, and perhaps also for the Pleistocene/Holocene boundary. Recent suggestions to abandon chronostratigraphic terms (system, series, stage) in favor of geochronologic terms (period, epoch, age) are logically defensible, but could perpetuate the continuing confusion between various stratigraphic categories.