The influence of residual cosolvent on the partitioning tracer technique for estimating a nonaqueous phase liquid (NAPL) saturation in porous media was investigated. Batch equilibrium and column miscible displacement tests were used to evaluate the influence of residual alcohol cosolvents in the aqueous phase on partitioning and transport of alcohol tracers through sandy soil columns containing tetrachloroethylene (PCE). As the volume fraction of cosolvent alcohol (f(c)) increased ( f(c) < or = 0.1; 10 vol %), partition coefficients (K(nc)) for the alcohol tracers linearly decreased for residual cosolvent ethanol, linearly increased for residual cosolvent tert-butyl alcohol, and did not exhibit an evident change for residual cosolvent 2-propanol. These observations are consistent with measured changes in solubility (S(c)) of the alcohol tracers over the same range (f(c) < or = 0.1) of these residual cosolvent alcohols. Column miscible displacement tests using ethanol as a residual cosolvent ( f(c) < or = 0.1) exhibited earlier partitioning tracer breakthrough leading to an underestimation of NAPL saturation (S(n)) when constant, cosolvent-free partitioning coefficients were assumed. The underestimation magnitude increased with higher initial residual cosolvent alcohol in the columns. The S(n) underestimates were not significant but were 1-10% lower than the actual S(n) (0.18). The estimated partition coefficients based on column tests with residual cosolvent (K(col)) were consistently less than those based on batch tests. Column tests with low (0.5%) and high (15%) S(n) revealed that the residual cosolvent alcohol effect was different depending on the amount of NAPL in the column. Using ethanol for a cosolvent (10%) and 2,4-dimethyl-3-pentanol as a partitioning tracer, the S(n) values were underestimated by about 17% and 5%, respectively, in the low and high NAPL saturation columns.