This document concerns three information fusion methods: Information Matrix Fusion (IMF), Covariance Intersection (CI), and Sampling Covariance Intersection (SCI). These methods are compared for performance under extreme multiple-counting conditions, that is, when an estimate is improperly fused to a track multiple times as if the estimate was repeatedly found by independent measurements. This situation can possibly occur in networked fusion systems where data pedigree is less than properly maintained, especially when an information relay is implemented to handle diminished communication environments. Extreme multiple-counting behavior in particular is examined for the purposes of this document. This research demonstrates that the normally preferable methods, IMF and SCI, are prone to falsely optimistic covariance values in such situations. All three fusion methods result in the state estimate approaching the estimate being repeatedly fused; the more conservative CI method also results in the covariance approaching that of the repeated estimate. We obtain these results through inference from the governing equations and examination of behavior under Monte Carlo simulations.