LNG spills on land are most likely to happen in dike/impoundment areas, introducing the need to consider effects of dike/impoundment walls on vapor holdup and dispersion. The DEGADIS and FEM3A models are approved for determination of vapor cloud exclusion zones in 49 CFR 193, but DEGADIS is limited to prediction of dispersion from area sources over flat, obstacle-free terrain and is therefore not directly applicable to spills in dike/impoundment areas. However, FEM3A, which is a computational fluid dynamics (CFD) model, is directly applicable. Before FEM3A was approved, ad hoc methods were widely used to determine input conditions for DEGADIS to estimate dispersion downwind of a dike system for such releases. Vapor holdup and mixing with air in a dike system have been studied in wind tunnel and field experiments, and the fallacy of using an overly simplistic estimation of the time and rate of vapor cloud overflow has been demonstrated. In some cases, these ad hoc methods have resulted in exclusion zone determinations that do not provide for public safety as intended by 49 CFR 193. This paper summarizes the current state of knowledge about this important vapor dispersion scenario, including key experimental data that address the question. Applications of FEM3A to determine exclusion zones for a hypothetical spill within a typical dike/tank configuration, as well as a shortcut use of FEM3A to describe vapor overflow from the same impoundment for use as input information for DEGADIS simulation of the dispersion downwind from the dike, are illustrated. © 2005 American Institute of Chemical Engineers Process Saf Prog, 2005