The purpose of this investigation was to develop and validate at full-scale dynamic models and control strategies for the activated sludge process capable of predicting both the clarification and thickening functions of the solids-liquid separator. This also included the development of a hydraulic model capable of predicting flow transients through the treatment plant and a solids model to predict MLSS concentrations in the reactors for arbitrary hydraulic forcings. These models were then utilized to derive control strategies to minimize the discharge of effluent suspended solids. A mixing model was identified using tracer tests. Analysis of the data demonstrates that a tanks-in-series model describes the reactor system better than a dispersion model. A hydraulic model was developed from mass balances and well known flow equations. The model demonstrates the limited dampening capacity of treatment plants for hydraulic disturbances. Full scale experiments to identify models and estimate parameters were performed at a 5 MGD wastewater treatment facility in Houston, Texas. A distributed computer monitoring and control system consisting of on-line instruments, programmable controllers, and a minicomputer were installed at the plant. A table-driven data acquisition and control software package was implemented. Numerous experiments demonstrated that influent flow rate and pattern were the most important factors affecting the effluent suspended solids concentration at the plant studied. Hydraulic transients had an immediate effect which persisted longer than the actual disturbance. The recycle flow rate had relatively small effects. The sludge blanket level also had little effect until it was very near the water surface. A model was proposed which incorporates these features. A sludge thickening model developed by Stenstrom was modified to account for the conical bottom of the settler. The model is capable of predicting the return sludge concentration and the accumulation of solids in the settler. Settling parameters were estimated from batch settling tests in a stirred vessel. These models were utilized to derive an influent pumping strategy to minimize the discharge of effluent suspended solids. The strategy employs flow forecasting and a Simplex optimization routine to utilize the dampening capacity of the wet well in an optimal manner. A recursive state/parameter estimation technique was adapted for use with the clarification and thickening models. This technique can be used to give better estimates of the model states and update the model parameters from on-line measurements.