The past decade has witnessed a remarkable advance in our understanding of globular cluster (GC) systems. Optical and ultraviolet imaging from HST has been used to study GC systems belonging to nearly one hundred (mostly early type) galaxies. Spectra from 4-10m class telescopes have yielded radial velocities for more than a thousand extragalactic GCs, along with metallicities for subset of the brighter objects. Infrared photometry for extragalactic GCs is appearing in the literature at an ever-increasing rate, promising to refine our understanding of how these systems formed and evolved.
Indeed, theorists are now hard pressed to produce a single model for the formation of GC systems that explains - or, at least, is consistent with - the enormous amount of observational material that has been accumulated. This is a daunting task, since any viable formation model must reproduce and explain the apparently constant number of GCs per unit baryonic mass, the striking complexity of GC metallicity distribution functions (MDFs), the observed spatial distributions of GCs about their host galaxy, and the emerging constraints on the ages and kinematics of the metal-poor (MP) and metal-rich (MR) GC subpopulations. What is ultimately required is a model that explains not just these observational properties, but also their variations (if any) with host galaxy environment and morphology. Although several important issues remain unresolved, I will argue below that there may be an emerging consensus on the origin of the MP GC subpopulations associated with bright galaxies. The origin of the MR subpopulation remains very much open to debate.