The use of Statistical Energy Analysis (SEA) in the field of vehicle noise is discussed. Theoretical fundamentals and basic assumptions of the method are summarized. Examples of successful prediction of interior noise levels in vehicles using the “classical” formulation for SEA are reviewed.Recently methods have been presented for the in-situ experimental determination of coupling- and internal loss factors for vehicles, based on the power balance equations. The methods are a result of applying the SEA hypothesis to multi-subsystem models of complex structures. This approach is attractive for vibratory power flow models of very complex structures such as car bodies. Simple substructures or junctions can not easily be identified for such structures why models based on theoretical estimations for basic substructures or junctions become uncertain. This experimental approach is sometimes called “inverse SEA”, since the elements of the loss-factor matrix are determined instead of the subsystem energies.Some serious fundamental problems and erroneous assumptions for the “inverse SEA” method are discussed and illustrated using a finite element and an analytical model of a simple plate structure.