Several characteristics of the isothermal thermoluminescence (ITL) signal from quartz grains held at 310 degrees C are investigated. The effect oil the TL signal of holding the quartz at this temperature for different periods of time is recorded. It is shown that the TL and ITL data cannot be explained by a single trap obeying non-first order kinetics. The pattern of TL signal loss implies that probably two TL peaks are removed, one by holding for similar to 5 s at 310 degrees C and the other by holding for 500 s. The ITL decay curves that result from holding the sample at 310 degrees C is analysed using the Sum of three first order decay functions, and these are termed fast, medium and slow ITL. Two signal integrals, 0-5 and 50-100s, are considered to be representative of the fast and medium ITL components. These are investigated with regard to their response to laboratory irradiation, thermal stability and the effects of optical bleaching. The (0-5s) signal shows all order of magnitude higher for saturation level with laboratory dose than the optically stimulated luminescence (OSL) signal, and has a thermal stability that is very sintilar to published values for the OSL or the 325 degrees C TL peak. Bleaching with a solar simulator reduces both parts of the ITL signal to a low, but non-zero level in a few thousand seconds. This result is supported by an equivalent dose estimate oil a modern river sample from Zambia of 11 +/- 2 Gy. This combination of low residual dose and high saturation level appears to be Suitable for dating samples that are demonstrated to be beyond the range of OSL dating. (c) 2007 Elsevier Ltd. All rights reserved.