Optically Stimulated Luminescence (OSL) dating based on the fast component is now widely used to establish chronologies of sedimentary deposits. Since this component saturates at a relatively low dose, the method is limited to the dating of Late Pleistocene quartz samples. Consequently, dating events beyond this limit is a key challenge. In this context the use of the TT-OSL signal which exhibits a high doses saturation level offers the potential to extend the age range of the method. Since the promising study of Wang et al. (2006b), in which a 780 ka age validated by palaeomagnetism data was reported for a Chinese loess sample, several studies focused on establishing a SAR TT-OSL dating protocol. They suggested applying different kinds of heat treatment at the end of a SAR cycle, as well as two normalisation procedures. The aim of this study was to test the reliability of these protocols for a loess-like Middle Pleistocene sample for which the expected De is known approximatively. We tested four published SAR protocols by implementing the dose recovery tests on artificially bleached quartz aliquots. The results obtained showed a systematic overestimation of the recovered doses and revealed high sensitivity changes between the first (natural) and the following cycles. It is believed that this behaviour is a consequence of the thermal treatment applied at the end of each SAR cycle which is necessary to empty the hard-to-bleach TT-OSL traps. Neither OSL nor TT-OSL signal test dose responses proved suitable to correct this sensitivity change.