Over the last years, SEA has been recognized as a useful tool to model and analyze the high-frequency vibro-acoustic behavior of fully assembled complex structures. This paper discusses the experimental derivation of the loss factor model of a passenger car. The paper outlines the different steps which need to be taken to obtained a fully validated experimental SEA model. This includes the subdivision into subsystems, the PIM measurement campaign, the derivation of the loss factors and their associated confidence levels and the model validation. The paper further details how the experimental SEA model was used to quantify and investigate the airborne and structure-borne contributions to the interior noise level for a road noise test condition. The operational power inputs to the vehicle were indirectly determined from operational response measurements. A contribution analysis showed that airborne noise sources dominated structure-borne noise sources above 500Hz. A sensitivity study allowed to propose an effective design modification. This modification was implemented and successfully tested on the road.