Buccal cells serve as targets to assess oral exposures. We have refined isolation methods to characterise yield, viabilities, types of cells and baseline levels of genetic damage. Buccal cells were isolated from mouthwashes of 27 volunteers. They were characterised microscopically and different methods (using antibody-labelled magnetic beads, filtration and gradient centrifugation) were compared to separate epithelial cells from leucocytes. Viability of cells, DNA damage, and activity of glutathione S-transferase (GST) were measured with dye exclusion, microgelelectrophoresis, and biochemically. Mouthwashes contained approximately equal amounts of epithelial cells and leucocytes with detectable GST-activities. Repetitive determinations with mouthwashes from four individuals yielded per sample (3.5+/-1.4)x10(6) epithelial cells and (4.7+/-3.9)x10(6) leucocytes with viabilities of 8 and 94%, respectively. Epithelial cells could not be isolated using antibody-labelled beads, but cell separation with the leukocyte-specific antibody CD45 succeeded, yielding 37% leucocytes with a purity of 95% and viability of 65%. Filtering the mouthwash through a 10 microm filter yielded 57% leucocytes, with 86% purity and 94% viability. When using density gradient centrifugation as the separation method, the recovery of leucocytes was low (22%), but good results were scored for purity (95%) and cell viability (88%). This method was used to isolate leucocytes, which were then subjected to a micro-scale comet assay-modification. It was found that buccal leucocytes obtained from smokers had more DNA damage than cells from non-smokers. In conclusion, suspensions of buccal cells consist in approximately equal parts of epithelial cells and leucocytes. Only leucocytes are sufficiently viable for measuring parameters of cytotoxicity and genotoxicity or for studying modulation of gene expression. The cells are useful targets of non-invasive biomarkers, which could be incorporated as tools in many types of intervention studies.