Dating of quartz by optically stimulated luminescence (OSL) has been revolutionized with introduction of the test dose (TD) in development of a measurement sequence known as the single-aliquot regenerative-dose (SAR), whereby a valid sensitivity correction for the luminescence signal is provided in the measurement cycle. However, the size of the TD used in the SAR protocol remains controversial. Previous studies show that the TD has little effect on the equivalent dose (D_e) for young samples in luminescence dating in which the applicability of different deposits varies greatly in different regions. However, detailed studies are lacking on how TD size affects SAR–OSL results of samples with a relatively high D_e range. In this study, typical loess samples with high D_e values (∼60 Gy–∼250 Gy) from the eastern Tibetan Plateau were selected to investigate the effects of variation in TD size on the quartz SAR–OSL protocol. Dose recovery tests show that a known dose could be recovered successfully by applying different TDs. Test dose size has an effect on shapes of regenerated dose–response curves (DRCs) and has different influences on D_es and characteristic saturation doses for quartz samples with a high dose range. A TD size of 20%–30% D_e is a good compromise for Tibetan loess with D_e of ∼60–120 Gy in the quartz SAR protocol, and a TD size larger than 30% should be considered for samples with a larger D_e. The results of this study highlight the importance of TD size in the SAR–OSL protocol for quartz samples with a high dose range.