Current innovations in the area of power train engineering for cars are generating significant and unprecedented challenges for the development of acoustically unobtrusive transmissions. As regards relevance for customers, noise phenomena caused by loose parts play a key role. Transmission rattle, whose primary trigger is the pulse-like torque output of conventional combustion engines, also falls into this category.Current trends within the engine development (e.g. down speeding and downsizing) generate significant challenges for the development of acoustically unobtrusive transmissions. The aim of the presented approach is the evaluation and optimization of transmissions regarding the gear rattling noise perceived by the customer/passenger inside the vehicle. The investigation is divided into several experimental parts. The sound characteristics of the perceivable rattling noise is determined and evaluated as well as the transfer of rattling noise into the passenger compartment. Experiments are carried out with passenger cars on a dynamometer test rig. Several vehicle and transmission parameters are recorded while operating the vehicle at different speed and load conditions. At the same time binaural sound measurements are taken of the interior and exterior noise. The sound measurements are used for subjective rating by test persons in a sound laboratory regarding the perceivable rattling noise. Using a regression analysis for the comparison of the subjective ratings and the measured transmission and vehicle parameters, significant parameters are identified that show good correlation with the average subjective rating. Therefore objective ratings can be given to several vehicle-drivetrain combinations at different driving conditions. Furthermore critical levels and perception thresholds of these parameters are defined.For component testing a test rig is used which allows operating the transmission at boundary conditions comparable to the full vehicle. The input of the transmission is a non-uniform rotation of the input shaft which is identical to the transmission input in the full vehicle under the previously set operating conditions. Therefore, during the experiments on component level the previously identified significant parameters can be recorded and directly compared to the full vehicle measurements. On component level optimizations of the transmission can be carried out and verified by shifting the significant parameter below the critical level. Since the boundary conditions are identical to the full vehicle the expected customer rating can be improved. Hence, using this approach transmissions can be evaluated and optimized regarding the rattling behaviour in the full vehicle and the perception of the customer before reaching prototype stage.