A review of the information necessary to produce an operational fallout model. Each facet of the problem is discussed, the attendant difficulties pointed out, and the possible compromises suggested. A comparison of current fallout models is attempted so that strengths and weaknesses of each model can be assessed. Finally, a suggested program for the development of a flexible computer model is made. Such a model could provide a method for examining the variation in prediction as a function of the variation of the input.