With rising ambient temperature and atmospheric carbon dioxide levels, there is an urgent need to monitor soil carbon stocks over large regions of the earth. Near-infrared diffuse reflectance sensing (NIRS) of soils, using satellite- or airplane-based instruments, is increasingly regarded as a potential method of choice for this purpose. Considerable research has been devoted to NIRS in the last few years, but this research has been generally restricted to sieved air-dried soils analyzed under laboratory conditions. For NIRS to be useful for the estimation of soil carbon stocks in the field, a technique must be developed to account, among other things, for the presence of moisture in the surface layer of soils. In this context, a first objective of the research described in this article was to determine whether, for three soils with contrasting characteristics, a simple constant proportionality factor relates NIR spectra obtained at different moisture contents, and whether there is relative constancy of this proportionality factor among soils, suggesting the possibility of a practical strategy to correct NIR spectra for soil moisture. A second objective of the research was to use ratio and derivative analysis to identify portions of NIR spectra that appear least affected by moisture content and on which a determination of other parameters such as organic matter content could be based. Because constant proportionality of the spectra at different moisture contents seems elusive, at best, the most significant result obtained is the identification of specific wavelength ranges in the NIR spectra, at 800 to 1400 nm, 1600 to 1700 nm, 2100 to 2200 nm, and 2300 to 2500 nm, where the first derivative of the spectra seems independent of the moisture content of the soil samples. This observation suggests that an operational method could be developed, focused on these wavelength intervals, to obtain moisture-independent estimates of a range of soil parameters under field conditions.