Statistical energy analysis is generally used to study the vibroacoustic response of systems with high modal densities. The most accurate predictions are obtained at high frequencies where the modal overlap is high and many modes contribute to the response in each frequency band. Under these conditions, the vibrational response is fairly uniformly distributed over frequency and over the spatial extent of the SEA subsystems. Validation of an SEA model at high frequencies can be accomplished by comparing the predictions of average subsystem response with an average formed from measured data at a relatively small number of locations. At lower frequencies, where the modal overlap is not high, the vibrational response shows significant variability over both frequency and location. Large variability makes validation of the models more difficult. An accurate measurement of the average subsystem response requires that data be collected at a large number of locations and over wide frequency bands. The introduction of “modal power” as a basis of comparison helps to reduce the response variability and thereby allows SEA models to be validated using data from a smaller number of measurement locations and in narrower bands of frequency. In this paper modal power is defined in terms of basic impedance and modal density parameters. Modal power measurements are then compared with SEA predictions to show the reduced variability of the response and to demonstrate a method of SEA model validation.