During the period from 1977 to 1989, the Lawrence Livermore National Laboratory (LLNL) conducted a liquefied gaseous fuels spill effects program under the sponsorship of the US Department of Energy, Department of Transportation, Gas Research Institute and others. The goal of this program was to develop and validate tools that could be used to predict the effects of a large liquefied gas spill through the execution of large scale field experiments and the development of computer models to make predictions for conditions under which tests could not be performed. Over the course of the program, three series of LNG spill experiments were performed to study cloud formation, dispersion, combustion and rapid phase transition (RPT) explosions. The purpose of this paper is to provide an overview of this program, the lessons learned from 12 years of research as well as some recommendations for the future. The general conclusion from this program is that cold, dense gas related phenomena can dominate the dispersion of a large volume, high release rate spill of LNG especially under low ambient wind speed and stable atmospheric conditions, and therefore, it is necessary to include a detailed and validated description of these phenomena in computer models to adequately predict the consequences of a release. Specific conclusions include: * LNG vapor clouds are lower and wider than trace gas clouds and tend to follow the downhill slope of terrain due to dampened vertical turbulence and gravity flow within the cloud. Under low wind speed, stable atmospheric conditions, a bifurcated, two lobed structure develops. * Navier-Stokes models provide the most complete description of LNG dispersion, while more highly parameterized Lagrangian models were found to be well suited to emergency response applications. * The measured heat flux from LNG vapor cloud burns exceeded levels necessary for third degree burns and were large enough to ignite most flammable materials. * RPTs are of two types, source generated and enrichment generated, and were observed to increase the burn area by a factor of two and to extend the downwind burn distance by 65%. Additional large scale experiments and model development are recommended.