Root density responses to pasture grassland management treatments are time-consuming and expensive to measure with conventional root coring and washing procedures. We evaluated the use of laboratory-based near-infrared reflectance spectroscopy (NIRS) for assessment of proportion of pasture root dry mass in soil. Silt loam soils supporting mixed cool-season perennial grass-legume-forb mixtures under differing defoliation severity and frequency were sampled throughout growing seasons of 2012-2013. Cores to a depth of 15 cm were sectioned into horizons 2.5 cm thick and 7.6 cm in diameter. Spectra of core horizons were obtained over a range of 1250-2350 nm with a scanning monochromator with sample cups in rotating top window orientation (Unity Scientific) from a) scans of tops and bottoms of intact horizons at field wetness; b) coarsely-crumbled soil aggregates at field wetness; and c) finely-pulverized soil following 24 hr of air-drying. Roots were then separated by washing samples through a series of seives with a final hole diameter of 0.3 mm, then dried to constant weight at 60 C and expressed as weight proportions of dry core samples. Root density ranged from 0.01 to 1.03 g/100 g soil and averaged 0.14 g/100 g soil. Chemometrics software (Ucal, Unity Scientific) was used to develop NIRS prediction equations of root density as a function of spectral characteristics. The proportion of variation in root density that was explained by spectral variation was in decreasing order of intact cores, coarse aggregates, and air-dry pulverized samples, with values of 0.52, 0.42, and 0.15, respectively. Corresponding standard errors of cross-validation were 0.06, 0.07, and 0.07 g roots/100 g soil for intact cores, coarse aggregates, and air-dry pulverized samples, respectively. In spite of large water peaks in intact cores and coarse aggregates at field wetness that could at least partially mask spectral information associated with roots, intact cores appear to be an optimum form among those tested. This may be related to greater uniformity of particle size than in the coarsely-aggregated and pulverized forms. The relatively higher precision with which root density can be predicted in intact cores suggests the feasibility of NIRS determination of root density with field spectrophotometers.