The aim of this work was to determine how in-situ portable XRF (pXRF) analysis of ores (in this study, nickel and iron) may be consistent with data obtained from well-prepared samples analysed by conventional XRF methods, and to investigate potential sources of error in field measurements due to different chemical composition, moisture content and particle size. For this study, 201 samples of saprolite and laterite, sieved at 125 μm, were selected for their Ni and Fe contents (from 0.119 to 5.225 % Ni and from 4.7 to 37.2 % Fe for saprolite samples and from 0.38 to 2.94 % Ni and from 21.5 to 55.55 % Fe for laterite samples), compressed to make powder pellets and used to calibrate the pXRF. It is shown that for these two elements, pXRF can correlate extremely well (r2 > 0.98, relative standard error <7 %) with conventional laboratory results when samples are suitably prepared (dried, pulverized and pelletized). Laboratory analysis of a few artificially wetted samples shows that both Ni and Fe estimated concentrations are directly proportional to moisture content. In the field, two experiments were conducted, one on an air-dried drill-core and the second directly on the pit face. The results indicate that because of sample heterogeneity, Ni and Fe contents are better estimated in the laterite layer than in the saprolite layer. Despite the many challenges associated with field conditions, pXRF can provide useful benefits for a number of applications in the mining sector.