This report has presented work done in two major areas of system identification, time domain, and frequency domain analysis. In time domain analysis, the parametric method of representing a system, was chosen because it lends itself more effectively to practical applications. Modeling methodologies for single-input/single-output and for multivariable systems were developed. An algorithm was developed for general applications using the recursive least-squares estimation technique of parameter estimation. The algorithm is very flexible in that the vector and matrix sizes are automatically adjusted within the program to account for the system delay in data used in the estimation. In each test where the actual output was compared to the estimated out-put, the results appeared to be quite favorable. The tests with known process time delay also yielded good model performance. The time domain method of system modeling has good noise rejection capability. The time domain method can be performed from data evident of normal process perturbations, whereas the frequency domain method requires that specific driving inputs be applied to the process. The recursive nature of the developed algorithm has the potential for real-time implementation. Adaptive control and process monitoring are candidate application. 17 refs., 25 figs.