Observations and model studies of large-scale buoyant plumes show three major types for the horizontal distribution of density. Type 1 represents the typical coastal freshwater plume of observations. The buoyant discharge turns right in the northern hemisphere (toward downshelf) under earth rotation at its source. Type 2 is common in many numerical model studies. Most of the buoyant water at the inlet turns left (upshelf) along the coast to form a continuously growing intrusion. Type 3 turns right but exhibits a massive anticyclonic bulge which grows with time; its coastal current is weak and carries a small fraction of the inlet fresh water or buoyancy flux. The great majority of observed coastal plumes and their appended coastal currents have been type 1, while models have most often produced type 2 or 3. To remedy this disparity, modelers have imposed an ambient shelf current in the downshelf (right hand) direction of sufficient strength to produce type 1 plumes. Field observations of the Delaware Coastal Current are presented. They show type 1 plumes occur even when the ambient shelf flow is upshelf. Imposing a downshelf ambient current is not, then, a generally applicable remedy to obtain a type 1 plume in model studies. A common element in the configuration of these models is use of the simple inlet, a rectangular breach in the coastal wall with uniform inflow water properties. An analytic treatment of the resulting flow near the coastal wall upshelf of the simple inlet predicts a steady intrusion of buoyant water that increases with depth of the coastal wall. Subsequent numerical experiments with this model configuration confirmed these predictions qualitatively. The simple inlet and the coastal wall are thus suspect. Three changes in model configuration yielded numerical model results that were type 1. The first was adoption of an idealized estuarine inlet in place of the simple inlet. The second was use of a greatly reduced coastal wall. The third was use of inlet channel angles less than normal. With all three of these alterations, upshelf intrusion was very weak, perhaps at a level that would be undetected in field observations. The flow was nearly steady state and no massive bulge was present, despite the absence of ambient shelf motion. Each of these three changes to configuration generated clear differences in the state of the buoyancy-driven coastal current well downshelf, despite use of the same bulk discharge properties such as total volume and freshwater fluxes. Using the results of observations, laboratory models, and numerical models, one may attempt to codify the natural and model settings or configurations that select which plume type occurs. Type 3 plumes seem the clearest to predict. They require weak or absent downshelf ambient current, nearly normal inlet channel angle, a steep coastal bottom slope, and water depths much greater than the typical depth of buoyant water. Type 2 plumes, in contrast, all appear to be shallow water phenomena. In addition they require weak or absent ambient downshelf current, inlet flow angles near normal, and a steep coastal bottom slope or vertical coastal wall. Type 1 plumes are the most common in field observations. They too are shallow water phenomena. Model configurations that favor them include absence of a significant coastal wall and use of more realistic inlet flow fields than the simple inlet.