A knowledge of atmospheric density and its variation at satellite altitudes is needed for operational support of several military activities. Atmospheric density data computed from the decay rates of more than 40 satellites orbited during a five-year period were analyzed during this investigation. Multiple regression equations were derived to specify the density at 10-km intervals at altitudes from 180 to 300 km as a function of solar activity, time of day, time of year, and combinations of these variables. Density values depicted by these regression equations were then compared with density data which had not been used during the derivation of the equations. The results of this investigation indicate that multiple regression analysis and the associated screening procedure can produce equations from which computed density values are in close agreement with observed density data. The results lead to a conclusion that no single density model can meet all needs. These results also indicate that characteristics of a density model should be selected to fit the purpose for which the model is intended. Four possible density models designed for different purposes are suggested. (Author).