jats:p The Statistical Energy Analysis (SEA) approach has largely been used in vibro-acoustic modelling to predict the averaged energy in coupled vibrating structures and acoustic cavities. The average is performed over an ensemble of nominally identical built-up systems where random responses are observed at high frequencies after excitation. Over the years, this approach has been extended to predict the energy variance employing the statistics of the Gaussian Orthogonal Ensemble, and numerical and experimental evidence has supported the predictions of the mean and variance of energy of coupled vibrating structures. However, little experimental evidence is found to validate the prediction of the variance of energy in coupled structural-acoustic systems. In this work, the mean and variance of energies predicted from a statistical energy analysis model have been validated with experimental measurements on a structural-acoustic system, comprised by a flat thin plate coupled to an enclosed acoustic volume. The structural system has been randomised by adding small masses on arbitrary positions on the plate, whereas the randomisation of the acoustic cavity is achieved by allocating rigid baffles in random positions within the acoustic volume. In general, good agreement is found between the predictions of the model and the experimental results. </jats:p>