An abundance of experimental data has repeatedly shown correlations between sorption of pollutant chemicals and sorptive characteristics such as aqueous solubility and molecular weight and properties of the sorbent such as total organic matter and hard carbon content. It has become increasingly evident, however, that none of these factors are good predictors of desorption. More importantly, they fail to explain why contaminated field soils retain pollutant chemicals for many years despite properties that suggest rapid transport and dissipation. Resolution of this apparent contradiction may lie in understanding the microstructural character of soil particles and its impact on retention. In the experiments described here, in situ diffuse reflectance spectroscopy (DRIFT) is used to investigate sorption and desorption processes for water vapor on both specimen clay minerals and soils. The clay mineral group includes three of the most common types found in soils: kaolinite, montmorillonite, and illite. The soils have developed in widely scattered regions of the country and are investigated both with and without treatment to remove organic matter (OM). DRIFT investigations were conducted on all of these sorbents in a controlled environment chamber using a flow of nitrogen with water at 50% RH: desorption was initiated by removing water from the flow stream. Bands indicative of sorbed water in both liquid and vapor states were present in spectra of all samples prior to the introduction of water vapor and throughout the sorption and desorption periods. With the clay minerals, the appearance of the vapor phase band was clearly correlated with surface charge characteristics. For all of the soils, treatment to remove OM resulted in lower starting values for sorbed water in both physical states but larger intensity increases for the liquid phase band over the sorption period. Spectral patterns of behavior of both bands during the combined sorption/desorption period show both similarities and differences to the behavior observed earlier in experiments conducted using 1,2-dichloroethane (DCA) as the sorptive molecule. The similarities derive from the fact that both water and DCA exist in liquid and vapor states under ambient conditions. Because its molecules are highly polar and exert strong intermolecular forces, however, water is a much more sensitive molecular probe of the sorption environment. Combined with our earlier porosimetry data, results of these studies support the concept that sorption and desorption are to a large extent a function of the microstructural array created by inorganic constituents. They furthermore suggest that natural OM within soil particles impacts both processes through a reduction in the available pore space and by moderating the effective charge of the surfaces that comprise the pores. The current experiments include a second sorption/desorption cycle to test an additional hypothesis presented in the earlier studies. An important aspect of these arrays, whether OM is present or not, is that much of the pore space involved is nanometer in scale and is thus highly subject to capillary forces. The condensation of vapor phase chemical to a liquid occurs spontaneously within pores of nanoporous dimension because of free energy differences between the bulk and sorbed states. As the liquid advances into the porous network, sorbed chemical in the vapor state must also be present at the liquid meniscus. Intermolecular forces that facilitate capillary flow of liquids both into and out of pores are not effective for molecules in the vapor state, which are thus retained for extended periods of time within the pore system. Subsequent condensation events, whether of pollutant chemical or water, provide additional forces that can drive those vapor state molecules further into pore regions where molecular displacement is even more difficult. The results support this hypothesis.