Several methods for the preconcentration of dissolved gold in natural waters were evaluated for use with graphite furnace atomic absorption spectroscopy (GFAAS) or inductively coupled plasma mass spectrometry (ICP-MS). An anion exchange method prior to GFAAS, and a solvent extraction method prior to ICP-MS both proved to have similar recoveries, Iow limits of detection (0.4 and 0.2 ng l-l, respectively, for a 2-l sample) and good reproducibility. Parallel analyses of particulate gold were also carried out by instrumental neutron activation analysis (IN&Q with detection limits between 0.04 and 0.5 ng I-' depending on the volume of filtered water, to provide a complete evaluation of transported gold. These methods were subsequently tested on stream, spring and adit water samples, to check the accuracy of methods for natural waters. Results obtained by ICP-MS and GFAAS on natural water samples were comparable. The maximum dissolved gold concentration was 3 ng I-', and particulate gold formed less than 50% of the total amount of gold transported. No significant variation was found in the dissolved gold content of one spring water sample monitored monthly over a one year period.