Industry needs rapid, accurate, and inexpensive methods for compositional analysis of biomass and biomass-derived materials. These methods when applied to on-line analysis can be used to assess feedstock quality at the plant gate and ultimately provide near real time process control and optimization capability. The purpose of this work was to demonstrate the ability to reliably measure corn stover composition using an on-line near-infrared (NIR) instrument in an industrially relevant process environment. We used a FOSS Direct-Light 5000 on-line NIR system to measure the composition of corn stover moving on a conveyor. A 3-factor, 2-level factorial experiment was undertaken to study the effect of sensor height (7.6 cm and 22.9 cm), particle size (0.6 cm and 3.2 cm), and number of sub-scans (16 and 64) on measurements of component (e.g., glucan, xylan, etc.) concentrations. We also performed a limited study to determine the effect of temperature on instrument performance. The results demonstrate that it is possible to measure corn stover composition with such a system. However, careful placement and monitoring is required to ensure accurate results are obtained as sensor height and to a lesser extent temperature were found to affect measurement of component concentrations. The technology can be used but the development of good calibrations will require access to a broad range of compositionally diverse samples.