Hydrophobic and volatile chemicals have proven to be difficult to dose in cell assays. Cosolvents are often needed to dissolve these chemicals in cell culture medium. Moreover, the free concentration of these chemicals in culture medium may diminish over time due to metabolism, evaporation, and nonspecific binding to well plate surfaces and serum constituents. The aim of this study was to develop a partition-controlled dosing system to maintain constant concentrations of benzo(a)pyrene, 1,2-dichlorobenzene, and 1,2,4-trichlorobenzene in an ethoxyresorufin-O-deethylase (EROD) assay and a cytotoxicity assay with the rainbow trout (Oncorhynchus mykiss) cell lines RTL-W1 and RTgill-W1. Polydimethylsiloxane (PDMS) sheets were loaded with test chemicals in a spiked methanol/water solution and placed in the wells, filled with culture medium, of a 24-well culture plate. Cells were grown on inserts and were subsequently added to the wells with the PDMS sheets. The system reached equilibrium within 24 h, even for the very hydrophobic chemical benzo(a)pyrene. The reservoir of test chemical in PDMS was large enough to compensate for the loss of >95% of the test chemical from the culture medium. The PDMS sheets maintained medium concentrations constant for >72 h. Nominal median effect concentrations (EC(50)) were 1.3-7.0 times lower in the partition-controlled dosing systems than in conventional assays spiked using dimethyl sulfoxide (DMSO) as a carrier solvent, thus indicating that the apparent sensitivity of the bioassay increased when controlled and constant exposure conditions could be assured. The EC(50) values of the test chemicals based on free concentrations were estimated in the partition-controlled dosing systems using measured PDMS-bare culture medium partition coefficients. Results indicated that 61, 70, and 99.8% of 1,2-diclorobenzene, 1,2,4-trichlorobenzene, and benzo(a)pyrene were bound to serum constituents in the culture medium.