The analytical variability encountered when sampling for platinum- group metals (PGM) was statistically evaluated during a 1983 study by the Bureau of Mines and the University of Alaska. Sampling procedures were designed to systematically incorporate commonly used reconnaissance exploration techniques under actual field conditions. Analytical variability was evaluated at each procedural step. Analytical data from each sample from a single deposit grouped relatively well around the calculated mean for that deposit. Standard deviations did not exceed 43 pct of the mean for any of the deposits; more commonly, the deviation was about 25 pct. Statistical treatment of the data by analysis of variance indicated the most important source of error is selection of the sample site. Two or more replicate samples appear necessary to adequately quantify the PGM content. Secondly, significant variation occurs during preparation of a pulverized split for fire assay. The optimum sample size was in the 11-lb range; larger samples did not appreciably reduce variability. Smaller samples generally indicated a higher degree of variability. Evaluation of the sample method suggested the common grab sample, if carefully taken, is a useful indicator of the general range of PGM concentration but potentially includes a considerable margin of error.