One of the common measures for river bank protection is the installation of riprap. There are several methods to design riprap appropriately, which are however generally limited to dumped medium size blocks. Nevertheless, an additional resistance against erosion can be achieved by individually placing blocks in one or several layers instead of dumping them arbitrarily. An experimental investigation has thus been performed to study the stability of large blocks which are compressed as a river bank protection. Tests were carried out including one layer of stones as well as two layers, to evaluate the influence of the riprap layering (e.g. riprap thickness) on the bank stability. The effect of the thickness on the stability of ripraps is investigated in a 10 m long and 1.2 m wide tilting flume, with a rough fixed bed. Riprap median particle size was D50 = 37 mm. Testing was conducted for channel longitudinal slopes 0.015 and 0.030 and riprap bank inclinations of 27, 31 and 35 degrees. The riprap was installed on the top of a wide grain size distribution filter. Supercritical flow conditions were considered, given the steep channel slope. The complete removal of the riprap in a section under a constant discharge was defined as the failure criterion. The riprap failure threshold discharge was determined based on the series of tests with duration of maximum 180 minutes. In each test, riprap transport rate was measured every minute while the stones were tracked by a video camera and collected in a sediment trap at the channel end. The time of total failure was defined by standard video-image processing techniques. A time based analysis of failure was performed and first results revealed that, for similar conditions, the second layer stabilizes the riprap significantly and delays the time for total failure. Nonetheless, transport rate was found to be increased in this latter situation.