Because of increasingly stringent demands on product quality, process measurements of high precision, accuracy, and reliability are required in many processes where process variables are traditionally difficult to measure. In many applications, several different measurements of certain key process variables are available from various sources, including on-line analyzers, lab analyses, inferential measurements, and models of all types. Because these individual sensors often differ significantly in terms of their static and dynamic characteristics, precision, accuracy, sampling rate, measurement time delay, etc., in practice, only one of them is utilized for process monitoring and control purposes; the other sensor measurements are ignored. However, significantly improved estimates of the process variables can be obtained by utilizing all of the available sensor information simultaneously, i.e., by combining (fusing) these dissimilar measurements into a single estimate in a robust and fault-tolerant fashion. There is evidence that such sensor fusion operations are integral to the operation of biological control systems, and it is from one of these systems (the blood pressure control system) that we have drawn inspiration for developing the technique discussed in this article. A sensor fusion approach is developed using stochastic systems theory (in particular, Kalman filtering). The advantages of using multiple, redundant sensors, as well as the potential improvements achievable by combining delayed or multirate sensor information with that of a single nominal sensor, are then quantifiled. Robustness is achieved by augmenting the nominal technique with failure detection, classification, and compensation schemes, based on statistical hypothesis testing. The performance of the techniques is demonstrated on an actual plant application in which improved feed composition estimates are obtained for an industrial ethylene copolymerization reactor.