Acoustic emission (AE) signals are recognized as complementary measures for detecting incipient faults and condition monitoring in rotary machinery due to their containment of sources of potential fault energy. However, determining the potential sources of faults cannot be easily realized due to the non-stationarity of AE signals. Available techniques that are capable of evoking instantaneous characteristics of a particular AE signal cannot optimally perform in a sense that there is no guarantee that these characteristics (hereinafter referred to as the “features”) remain constant when another AE signal is obtained from the system, albeit operating under the same machine condition at a different time instant. This paper provides a theoretical framework for developing a highly reliable classification and detection methodology for gas turbine condition monitoring based on AE signals. Mathematical results obtained in this paper are evaluated and validated by using actual gas turbines that are operating in power generating plants, to demonstrate the practicality and simplicity of our methodologies. Emphasis is given to acoustic emissions of similar brand and sized gas turbine turbomachinery under different health conditions and/or aging characteristics.