A trial of near-infrared reflectance spectroscopy (near-IRS) to predict fecal concentration of dietary fiber (DF) in humans was performed. Fecal samples from 34 persons consuming 6 diets of known DF content were scanned by near-IRS. Diets consisted of a liquid nutrition supplement and quick breads containing either 0 g of fiber, 10 g of wheat bran, 30 g of wheat bran, 10 g of vegetable fiber mix, 30 g of vegetable fiber mix, or 30 g of sugar beet fiber. Neutral detergent fiber (NDF) determinations were made on all fecal samples. An equation to predict NDF concentration was derived from 109 fecal samples using modified partial least-squares analysis. Neutral detergent fiber concentrations predicted from near-IRS data were compared to observed NDF concentrations of 75 other samples to validate the equation. Comparison of observed vs predicted NDF revealed nearly perfect correlation between the two. Calibration and validation of a near-IRS equation generally requires that tradiational chemical analysis be done on only two subsets of samples, thereby reducing the time and effort needed to analyze DF concentrations. This study demonstrates that near-IRS is a viable tool for studying DF digestibility in vivo in humans.