Measurements of ore particle composition distribution, commonly termed mineral liberation distribution, are used in assessing process performance in mineral processing. In many applications, comparisons are made between particle composition distributions (for example comparing the products of fine and coarse grinds) and in such comparisons it is useful to understand the errors in the measurements in order to decide whether any differences are significant. A statistical approach based on bootstrap resampling has been applied to estimate the confidence intervals for ore particle composition distribution measurements obtained using the MLA automated mineralogy system. In this approach confidence intervals for each individual composition class are estimated as compared to a previous analytical solution which provides this information for particle composition data in cumulative form (Leigh et al., 1993). The effects on the magnitude of the error associated with measured values of particle composition distribution of the number of ore particles measured in the analysis and the complexity of the particle texture are investigated. Examples from a gold-bearing pyrite ore and an iron oxide copper gold ore are presented to demonstrate the practical application of this approach. (C) 2015 Elsevier Ltd. All rights reserved.