Samples were collected from major strata in the upper 5 m of an alluvial soil profile in the Amargosa Desert of southern Nevada to compare rubber-balloon and drive-core bulk-density measurement methods. For strata where the fine soil was < 82% sand and < 15% clay, differences between total and fine-soil bulk-density values determined by the two methods were typically < 10 and 15%, respectively, even where rock-fragment content was as great as 48% by volume. Outside this range of fine-soil texture, where soil consistency was either very loose or very hard, the core method appeared to sample inaccurately, resulting in bulk-density values > 0.30 Mg m-3 less than those determined by the balloon method. Under the severe sampling conditions encountered, large decreases in the relative accuracy of the core method were not directly related to rock-fragment content, but were related to extremes in the cohesiveness of the strata sampled.