This thesis deals with several process control issues related to Syncrude's extraction process. Since startup in 1978, many changes have been made to the extraction control system and numerous new sensors were installed. The potential benefits of these changes and additions have not been fully realized as yet due to a lack of suitable real time models capable of integrating the information from the new sensors into the overall control strategy. The topics discussed here are part of an ongoing effort by Syncrude Research to improve the control strategy for the extraction plant to achieve better reliability and throughput for the process. These include the development of a method for on-line calibration and performance evaluation of composition analyzers applied to the plant feed composition analyzer, spectral analysis of process data and an example for the design and implementation of a suitable filter to improve control loop performance, process time delay estimation and empirical model identification.