The loess–palaeosol sequence of Batajnica (Vojvodina region, Serbia) is considered as one of the most complete and thickest terrestrial palaeoclimate archives for the Middle and Late Pleistocene. In order to achieve a numerical chronology for this profile, four sets of ages were obtained on 18 individual samples. Equivalent doses were determined using the SAR protocol on fine (4–11 μm) and coarse (63–90 μm) quartz fractions, as well as on polymineral fine grains by using two elevated temperature infrared stimulation methods, pIRIR290 and pIRIR225. We show that the upper age limit of coarse quartz OSL and polymineral pIRIR290 and pIRIR225 techniques is restricted to the Last Glacial/Interglacial cycle due to the field saturation of the natural signals. Luminescence ages on coarse quartz, pIRIR225 and pIRIR290 polymineral fine grains are in general agreement. Fine quartz ages are systematically lower than the coarse quartz and pIRIR ages, the degree of underestimation increasing with age. Comparison between natural and laboratory dose response curves indicate the age range over which each protocol provides reliable ages. For fine and coarse quartz, the natural and laboratory dose response curves overlap up to ~150 and ~250 Gy, respectively, suggesting that the SAR protocol provides reliable ages up to c. 50 ka on fine quartz and c. 100 ka on coarse quartz. Using the pIRIR225 and pIRIR290 protocols, equivalent doses up to ~400 Gy can be determined, beyond which in the case of the former the natural dose response curve slightly overestimates the laboratory dose response curve. Our results suggest that the choice of the mineral and luminescence technique to be used for dating loess sediments should take into consideration the reported limited reliability.