Because of the methodological difficulties associated with the MTT assay in screening short-term cultures derived from human malignant glioma, a chemosensitivity assay based on the protein staining using sulforhodamine B (SRB) has been optimized for use with these cells. SRB at a fixed dye concentration achieved maximal staining density at 20 min for most cell lines and this intensity was not further increased by using dye concentrations above 0.2%. A delay in staining after fixation did not significantly decrease staining intensity, but delay in dye extraction after fixation and staining did. There was an excellent quantitative and qualitative linear relationship between cell number determined by either the SRB assay or by cell counting, but not with the MTT assay which consistently underestimated the number of cells in assay plates. The MTT assay appeared to be incapable of detecting less than about 150 cells/well, while these small numbers of cell were readily detectable by either cell counting or SRB staining. There was a close correlation between chemosensitivity values derived from the MTT and SRB assays for procarbazine, CCNU and vincristine when the endpoint is taken as either the ID25, ID50 or ID75. The results indicate that the SRB is capable of producing broadly similar results to the MTT assay, but is more sensitive in the detection of small numbers of cells with a linear relationship between cell number and SRB staining intensity over a wide range of cell numbers. It is capable of producing data from short-term cultures from malignant glioma and offers technical advantages over the MTT assay in that plates may safely be stored at certain points during the assay without the need for immediate processing. The SRB assay provides a useful alternative to the MTT assay for determining the sensitivity of short-term cultures of human glioma to cytotoxic drugs.