Thermoluminescence (TL) is routinely used to date heated lithic artefacts which mostly consist of silex (a mixture of amorphous opal and microcrystalline chalcedony). Analytical investigations of bulk samples confirmed that these materials contain considerable concentrations of radioactive elements, generating an internal dose rate contribution. Common dosimetric models assume the latter to be homogeneous throughout the sample. If this assumption would prove invalid, this will result in systematic errors in the calculated age, especially in the course of so called 'hot spots' of alpha-emitters (and associated local changes in a-sensitivity) and the dose response characteristics of alpha-radiation. Laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) analyses of 22 silex samples are presented here, quantifying element concentrations at several tens analytical spots per sample. Along with radioactive elements (K, Rb, U, Th), another 21 major, minor and trace elements were measured in order to allow characterization of the impurities present in most of the samples. The dataset provides a detailed picture of the spatial distribution of radionuclides and hence of the uniformity of the internal alpha- and beta-dose rate. It is shown that the silex itself mostly contains low amounts of K (<0.1 wt.%), U (<1.0 mu g g(-1)) and Th (<0.4 mu g g(-1)), and dosimetrically negligible Rb concentrations. Systematically higher concentrations are obtained by ICP-MS measurements of the bulk samples. This matches with the finding that impurities (veins, inclusions) often yield significantly elevated radionuclide concentrations, up to two orders of magnitude higher than the silex values. These veins and inclusions, for example Ca or Mg carbonates and Fe-Mn-oxy-hydroxides, lead to steep gradients mainly in the internal alpha-radiation field. Alternative approaches are required to account for the non-uniform internal dose rate and improve the reliability of TL dates of problematic samples.