The power and generality of the finite element approach to structural analysis is evident from the literature. However, this very generality demands that the user define his problem to the finite element computer program in great detail and with precision. To extend the use of this method within industry and to approach more complex dynamic, nonlinear problems requires new solutions to the problem of generating large, error-free piles of data and improvements in the efficiency of the computer programs and in the flexibility of model description. This paper describes near- and long-term developments aimed at solving these problems.