Recent technological improvements have led to the widespread adoption of field portable energy dispersive X-ray fluorescence (FP-XRF) by governmental agencies, environmental consultancies and research institutions. FP-XRF units often include analysis modes specifically designed for the quantification of trace elements in soils. Using these modes, X-ray tube based FP-XRF units can offer almost "point and shoot" ease of use and results comparable to those of laboratory based instruments. Nevertheless, FP-XRF analysis is sensitive to spectral interferences as well as physical and chemical matrix effects which can result in decreased precision and accuracy. In this study, an X-ray tube-based FP-XRF analyser was used to determine trace (low ppm) concentrations of As in a floodplain soil. The effect of different sample preparation and analysis conditions on precision and accuracy were systematically evaluated. We propose strategies to minimise sources of error and maximise data precision and accuracy, achieving in situ limits of detection and precision of 6.8 ppm and 14.4%RSD, respectively for arsenic. We demonstrate that soil moisture, even in relatively dry soils, dramatically affects analytical performance with a signal loss of 37% recorded for arsenic at 20 wt% soil moisture relative to dry soil. We also highlight the importance of the use of certified reference materials and independent measurement methods to ensure accurate correction of field values. (C) 2012 Elsevier B.V. All rights reserved.