There are different loudness calculation procedures, such as the German standard DIN 45631/A1 and the proposed international standard ISO 532-1 (both based on Zwicker's loudness model) as well as the Dynamic Loudness Model (DLM) (by Chalupper and Fastl), the Time Varying Loudness (TVL) model (by Glasberg and Moore), and the loudness calculation algorithm based on a hearing model of Sottek, allowing for the prediction of the perceived loudness of time-varying sounds in many cases. However, recent studies show that the predictions for some tonal and impulsive sounds do not match the loudness ratings of normal-hearing listeners. To address this issue, the influence of specific signal properties of the sounds on the assessment of loudness was examined focusing on impulsive sounds. On the basis of these experiments, it was studied how far an own improved modeling approach can account for the specific signal properties of these sounds.