Model-based processing is a method of enhancing processing performance (smaller error variance) by the inclusion of physical models of the signal/medium, noise, and measurement systems into the processor. It is based on a state-space approach, which leads to Kalman type estimators. A major advantage of this approach is that it allows for the stochastic aspects of the problem to be included in a natural and self-consistent manner, thereby allowing for modeling errors to be accommodated by the processor. In this work we describe the application of model-based processing to the problem of processing towed array data. Here the focus is on four problems of interest in the field of towed array processing: (1) the bearing and source frequency estimation problem for plane waves; (2) the bearing, source frequency, and range estimation problem for the circular wavefront signal model; (3) problems (1) and (2) generalized to the case of the Rayleigh fading channel; and (4) the case of the stochastic broadband signal model. Results based on simulated data will be shown which will demonstrate the improvement in performance over conventional array processing techniques. In particular, it will be shown how this approach provides a passive synthetic aperture effect in a natural way.