A commonly used technique for evaluating potential bioreductive drugs is the determination of hypoxic cytotoxicity ratios in vitro. This experimental model, however, does not accurately mimic the tumour microenvironment, as other factors (such as reduced pH, poor nutrient status, low cell proliferation rates and high catabolite concentrations) are not incorporated into the design of the assay. Plateau-phase monolayer cultures possess many of these characteristics, and this study compared the response of plateau-phase and exponentially growing human colon carcinoma cells (DLD-1) with a series of standard and bioreductive compounds. All drugs tested were added directly to conditioned medium and three patterns of chemosensitivity were observed. In the case of doxorubicin, vinblastine and 5-fluorouracil, exponentially growing cells were significantly more responsive than plateau-phase cultures. ThioTEPA and MeDZQ (2,5-diaziridinyl-1, 4-benzoquinone) were equally cytotoxic to both populations of cells. Tirapazamine (SR4233), RSU 1069, mitomycin C and EO-9, however, were preferentially toxic towards plateau-phase compared with exponentially growing cells. While the exact mechanisms responsible for these observations in each case are not known, this study suggests that plateau-phase cultures may prove to be a useful experimental model in the evaluation of drugs designed to work preferentially within the tumour microenvironment.