In this paper we consider the fusion of quasi-redundant sensors to calculate the best estimate of a measurand and provide a measure of confidence in the estimated value. The developed methodology integrates a concept of self-confidence of individual sensors and quasi-redundant sensors. The fusion algorithm utilizes a Parzen estimator for calculating a probability distribution function (PDF) for the measurand. The PDF is formed based on weighted Gaussian functions whose parameters depend on the sensors' average noise level and the self-confidence. The PDF is used to calculate a best estimate as well as a level of confidence in the estimate. The methodology for the calculation of the best estimate and confidence is demonstrated using experimental data obtained from a research iron-melting cupola furnace.