Models used for system monitoring must strike a balance between stability and elasticity. Ideally, a model should adapt to new operating conditions without losing the ability to differentiate faults from nominal conditions. To this end, an adaptive nonparametric model (ANPM) has been developed for integrated monitoring, diagnostic, and prognostic use on small to medium size reactors. This paper gives an overview of the development of the ANPM with two example applications. The ANPM's original intent is to adapt a nonparametric model's memory matrix from data created using a first principle model (FPM) to the system's actual unfaulted data. This would be useful for monitoring new system designs from first construction and operation when the only available data are from FPMs. The FPM's data are used to build the best possible models initially, but during the system's operation, new data can be collected that are more accurate for future empirical model predictions. The use of the ANPM is demonstrated on two systems. The first system is a heat exchanger model that is modeled in SIMULINK with both a low-fidelity and a high-fidelity simulation. The second system is a flow loop, a physical system at The University of Tennessee that is also modeled in SIMULINK. The results of testing the ANPM on nonfaulted conditions for the heat exchanger model and the flow loop are given. Areas of future work and development are outlined.