This paper describes, qualitatively and through visual observations, the ice-sheet interaction with a shoulder ice barrier (SIB) during model tests. The model tests were performed in the Hamburg Ship Model Basin (HSVA) during July 2007. Since the SIB represents a new concept in ice barrier structures, model tests were intended to evaluate the general performance of the SIB. The paper describes seven different experiments where the ice thickness, the ice flexural strength, and the shoulder angle of the SIB are the parameters which are varied among them. The results are presented in two sections. The first part refers to observations common to all the experiments, where the ice failure mode and shoulder performance are given special attention. The second part describes the phenomena observed in each particular experiment in more detail. The former analysis allows for the visual identification of three phases (as mentioned in previous publications) and gives a deeper insight into the characteristics of each phase. The latter analysis on the contrary, allows us to achieve interesting conclusions about the SIB performance under different ice conditions and with different shoulder inclinations. A comparison between the failure mode observed during the model tests and observations presented in the literature of full scale vertical and sloped structures, ice interaction with rubble accumulation, is performed. The similarities found in the study between the model and full scale observations lead us to assume that the observed model test behavior may be expected during ice-SIB interaction in full scale conditions. However, some events that could be associated with the problems of the model, such as ice scaling, are highlighted. As a conclusion regarding the SIB performance, it is shown that the shoulder section, which is the principal innovation of the concept, satisfactorily accomplishes its task and represents a key modification to traditional ice barriers in generating smaller ice pieces and avoiding ice overriding.