The complex mix of organic and inorganic components present in urban and rural soils and sediments potentially enable them to provide highly distinctive trace evidence in both criminal and environmental forensic investigations. Organic components might include macroscopic or microscopic plants and animals, pollen, spores, marker molecules, etc. Inorganic components comprise naturally derived minerals, mineralloids and man-made materials which may also have been manufactured from mineral components. Ideally, in any forensic investigation there is a need to gather as much data as possible from a sample but this will be constrained by a range of factors, commonly the most significant of which is sample size. Indeed, there are a very wide range of analytical approaches possible, and a range of parameters that can be measured in the examination of the inorganic components present in a soil or sediment. These may include bulk colour, particle size distribution, pH, bulk chemistry, mineralogy, mineral chemistry, isotope geochemistry, micropalaeontology and mineral surface texture, amongst a host of others. Whilst it would be prudent to utilise as many parameters as possible, the forensic significance of the resultant data should be carefully considered. In particular, many parameters that could be measured (such as particle size distribution, pH and colour) may be affected by the nature of the sample, and vary temporally, or in response to sample storage conditions and hence, provide misleading results. The inorganic components of soil and sediment are typically relatively inert and therefore, potentially, one of the more valuable parameters to measure. To this end, we have utilised an automated scanning electron microscope with linked energy dispersive X-ray spectrometers (QEMSCAN®) in numerous criminal forensic investigations. The sample measurement is operator independent and enables detailed characterisation of the mineralogy of even small samples. The reproducibility of automated SEM-EDS analysis using QEMSCAN was tested by the repeated analysis of the same samples using the same operating, measurement and processing parameters. The results show that instrumental variability is significantly less than observed natural variability between samples. The observed instrumental variability is interpreted to be a function of the sub-sampling of a different sub-set of the overall assemblage of particles present in the samples analysed. Thus automated mineral analysis can be regarded as a robust, highly repeatable method in forensic soil examination.