This work assesses the prediction capability of COSMO-RS for solvent selection in dispersive liquid phase microextraction systems. The assessment is performed by comparing experimental partition ratios from previous works and calculated ones for different systems, including pesticides, herbicides, polyaromatic aromatic hydrocarbons, and halogen-containing pollutants from water samples. Results indicate that COSMO-RS mainly overestimates the partition ratios. Despite this, a priori analysis was carried out for conventional solvents used as extractants. Results showed that for pesticides and herbicides, COSMORS predictions agree just for a few extractants. For polyaromatic hydrocarbons (PAHs), the ring structure is a crucial parameter in extractants’ selection, and for Halogenated pollutants, the halogen content and kind of substitution are crucial to correct trend prediction. Besides this, ILs also were studied, finding that COSMO-RS overestimates the use of ILs, and higher deviations are found for longer alkyl chain cations. However, the Dichlorodiphenyltrichloroethane (DDT) metabolite extraction with ILs is well predicted. Dispersants were also evaluated, but depending on the kind of pollutant, different deviations were found.