Predicting the fate of organic compounds in streams and rivers often requires knowledge of the volatilization characteristics of the compounds. The reference-substance concept, involving laboratory-determined ratios of the liquid-film coefficients for volatilization of the organic compounds to the liquid-film coefficient for oxygen absorption, is used to predict liquid-film coefficients for streams and rivers. In the absence of experimental data, two procedures have been used for estimating these liquid-film coefficient ratios. These procedures, based on the molecular-diffusion coefficient and on the molecular weight, have been widely used but never extensively evaluated. Liquid-film coefficients for the volatilization of benzene and eight alkyl-substituted benzene compounds (toluene through n-octylbenzene) from water were measured in a constant-temperature, stirred water bath. Liquid-film coefficients for oxygen absorption were measured simultaneously. A range of water mixing conditions was used with a water temperature of 298.2 K. The ratios of the liquid-film coefficients for volatilization to the liquid-film coefficient for oxygen absorption for all of the organic compounds were independent of mixing conditions in the water. Experimental ratios ranged from 0.606 for benzene to 0.357 for n-octylbenzene. The molecular-diffusion-coefficient procedure accurately predicted the ratios for ethylbenzene through n-pentylbenzene with a power dependence of 0.566 on the molecular-diffusion coefficient, in agreement with published values. Predicted ratios for benzene and toluene were slightly larger than the experimental ratios. These differences were attributed to possible interactions between the molecules of these compounds and the water molecules and to benzene-benzene interactions that form dimers. Because these interactions also are likely to occur in natural waters, it was concluded that the experimental ratios are more correct than the predicted ratios for application purposes in the reference-substance concept. Predicted ratios for n-hexylbenzene, n-heptylbenzene, and n-octylbenzene were larger than the experimental ratios. These differences were attributed to a sorption-desorption process between these compounds and the surfaces of the constant-temperature water bath. Other experimental problems associated with preparing water solutions of these slightly soluble compounds also may have contributed to the differences. Because these processes are not part of the true volatilization process, it was concluded that the predicted ratios for these three compounds are probably more correct than the experimental ratios for application purposes in the reference-substance concept. Any model of the fate of these compounds in streams and rivers would have to include terms accounting for sorption processes, however. The molecular-weight procedure accurately predicted the ratios for ethylbenzene through n-pentylbenzene, but only if the power dependence on the molecular weight was decreased from the commonly used -0.500 to -0.427. Deviations for the low- and high-molecular-weight compounds were similar to those observed for the molecular-diffusion-coefficient procedure.