Abstract Difficulties in the application to 40–60 μm Chinese loess samples of the SAR thermally transferred OSL (TT-OSL) protocol outlined in Wang et al. (Wang, X. L., Wintle, A. G., Lu, Y. C., 2007. Testing a single-aliquot protocol for recuperated OSL dating. Radiation measurements 42, 380–391) are reported. These include poor recycling, negative intercepts on the sensitivity corrected TT-OSL axis and non-linear relationships between regeneration dose TT-OSL and test dose OSL that do not pass through the origin. A modified protocol is developed that attempts to circumvent these complications. This protocol involves correction for sensitivity change through the use of the TT-OSL response to a test dose and requires no correction for either charge carry over or basic transfer. A high temperature blue light bleach (400 s at 280 °C) is used in the middle and at the end of each SAR cycle to remove any TT-OSL signal remaining after previous dosing. The protocol appears to be applicable to a number of samples, producing reproducible dose response curves that within errors pass through the origin and saturate at high doses; the TT-OSL response to a test dose appears to be a satisfactory monitor of sensitivity changes. Testing the protocol on a Chinese loess sample shows that there is still signal growth up to a dose of at least 12 kGy. Dose recovery tests are also successful on a variety of samples and can recover known doses up to between 0.56 and 1.2 kGy. Reproducible growth is also observed using a number of coarse grained samples from various depositional environments and locations. However, signal strength is a limiting factor and many samples do not show sufficient TT-OSL sensitivity for application with any TT-OSL protocol.