We have studied various factors involved in the optimal use of a tetrazolium (MTT) based colorimetric assay for cell growth and chemosensitivity. The assay is dependent on the ability of viable cells to metabolise a water-soluble tetrazolium salt into a water-insoluble formazan product. We have found that DMSO is the best solvent for dissolving the formazan product, especially where a significant amount of residual medium is left in the wells of the microtitre tray used for the assay. A reaction occurs between medium and a solution of MTT formazan in DMSO which changes the shape of the absorbance spectrum of the solution. The resulting optical density is not however greatly dependent upon the volume of added medium in the range 1-10 microliters. Between 10 and 40 microliters of added medium results in a gradually lower optical density than that produced by the smaller volumes. Above 40 microliters, the optical density increases again due to turbidity as protein precipitation occurs. When cells are incubated with MTT, the resulting optical density of the formazan product is dependent upon both the concentration of MTT and the incubation time. The optical density is stable for several hours after solution of the formazan in DMSO. A linear relationship is seen between optical density and cell number for incubation times of 2, 4, 6 or 24 h with 20 microliters of MTT (5 mg ml-1) added to 200 microliters medium. We have adopted 4 h as the standard incubation time for the assay. Only a small amount of MTT formazan product can be detected in the growth medium of wells in which cells have been exposed to MTT. Comparative chemosensitivity data for EMT6 mouse tumour cells show good agreement between results obtained using the MTT assay and results based on total cell count after a fixed period of growth.