In optical dating of feldspars, it is commonly supposed that potassium (K)-rich feldspar grains have been isolated for luminescence measurements by means of density separation. The K and rubidium (Rb) concentrations of these grains deliver a significant internal beta dose rate. K and Rb concentrations of 12.5 ± 0.5 wt% and 400 ± 100 ppm, respectively, are typically assumed for dose rate and age calculation, but this assumption may be wrong. Thorium (Th) and uranium (U) also contribute to the beta and alpha components of the internal dose rate. We investigated the K, Rb, Th and U concentrations of individual luminescent feldspar grains in four sediment samples from Liang Bua in Indonesia to determine single-grain internal dose rates and appropriate sample-average internal dose rates. The composition of the ‘K-rich’ feldspar fraction was investigated using X-ray diffraction spectrometry (XRD), which found that this fraction was heavily contaminated in each sample by quartz and plagioclase feldspars. Grains with detectable post-infrared infrared stimulated luminescence (pIRIR) signals were characterised using a range of techniques, including energy-dispersive spectroscopy (EDS), wavelength-dispersive spectroscopy (WDS) and quantitative evaluation of minerals using EDS (QEM-EDS). The grains were found to be composed of a range of minerals, including feldspars, quartz, clay minerals and heavy minerals, as well as volcanic glass. The K concentrations of these composite grains were determined using QEM-EDS, and their Rb, Th and U concentrations were determined using laser ablation inductively-coupled plasma mass spectrometry (LA-ICP-MS). The calculated internal dose rates were very low for the majority of grains in each sample (<0.2 Gy.ka−1 for 180–212 μm-diameter grains), although a wide spread of internal dose rates was also observed. None of the samples showed any relationship between internal dose rate and either pIRIR signal intensity or equivalent dose (De). Two samples of clastic sediment were found to have different distributions of internal dose rate to those of two samples of tephra. To obtain reliable ages for sediment samples from Liang Bua thus requires the use of appropriate whole-of-grain and sample-average internal dose rates for grains with composite mineralogies deposited in these contrasting sedimentary contexts. More generally, greater consideration needs to be given to the internal dose rates of feldspar grains in volcanic regions such as Island Southeast Asia, especially composite grains with complex mineralogies, to ensure that the calculated ages are accurate.