This research was concerned with the analysis and interpretation of water quality data in order to develop improved mathematical models for predicting water quality. Spectral analysis was employed to analyze water quality data from the Coosa, Detroit, Missouri and Ohio rivers. Mathematical models were developed to represent the processes taking place in the stream. Parameter estimation techniques were used to determine the value of the constants and the parameters in the model equations. The study also included (1) estimation of missing values in water quality data, (2) experimental simulation of thermal discharge monitoring and application of spectral analysis techniques, (3) compilation of water quality data and parameters from Kansas rivers and streams, (4) identification of parameters in transient water quality models from stochastic data, and (5) evaluation of methods for estimating stream water quality parameters in a transient model from stochastic data. (Modified author abstract)