This study established the preferences of shrubsteppe granivores among seeds of 6 common sagebrushsteppe plants and related the preferences observed to physical and nutritional attributes of the seeds. Seeds of big sagebrush (Artemisia tridentata), cheatgrass (Bromus tectorum), Indian ricegrass (Oryzopsis hymenoides), western wheatgrass (Pascopyrum smithii), bitterbrush (Purshia tridentata) and green needlegrass (Stipa viridula) were placed in groups of petri dishes designed such that seed removal could be ascribed to either diurnal vertebrates, nocturnal vertebrates or ants. Though absolute quantities of seeds removed varied among the 3 granivore classes, calculations of preference based on weights of each seed species removed by each granivore class indicated that all 3 ranked the seeds similarly. Preference hierarchies of the 3 granivore classes were highly positively correlated with both calories per seed and % soluble carbohydrate of the seeds. The first correlation supports a basic prediction of optimal foraging theory -that foragers should maximize energy intake per unit time spent foraging. Both correlations emphasize the role of seed nutritional qualities in granivore seed selectivity in that soluble carbohydrate is a water-efficient energy source and its percentage is a good indicator of the digestible energy available in a food item. A corollary experiment comparing granivore use of an exotic seed (millet [Panicum miliaceum]) and a preferred native seed (Oryzopsis) demonstrated a distinct preference for the exotic. Since millet seeds are particularly high in % soluble carbohydrate, this result reinforced the apparent value of this nutritional attribute as a predictor of granivore seed preference. Among many seed resource characteristics upon which granivore seed selectivity might operate, our results indicate that individual species' nutritional composition may be particularly important. Thus, inferences about seed selectivity and resource partitioning among arid-land granivores should be interpreted with caution, especially those based on experiments using seed introductions, since the influence of seed nutritional attributes has not been widely acknowledged heretofore.