The European Standards Committee on Oxidative DNA Damage (ESCODD) recommended the use of the lesion-specific repair enzyme, formamidopyrimidine DNA-glycosylase (FPG) in the comet assay to detect oxidative DNA damage. In the present study, FPG was compared with endonuclease III (ENDOIII) and human 8-hydroxyguanine DNA-glycosylase (hOGG1) for the ability to modify the sensitivity of the comet assay. Mouse lymphoma L5178Y cells were treated with dimethyl sulphoxide (DMSO) as a standard solvent or reference agents known to induce oxidative damage (gamma irradiation and potassium bromate) or alkylation (methyl methanesulfonate, MMS; ethylnitrosurea, ENU). Using DMSO even up to toxic concentrations, no increase in breaks was seen with FPG, ENDOIII or hOGG1. With gamma irradiation (1-10 Gy), dose-related increases in breaks were seen with all three enzymes. FPG and hOGG1 gave similar increases in breaks after potassium bromate treatment between 0.25 and 2.5 mmol/l, but ENDOIII showed an increase only at the highest concentration, 2.5 mmol/l. Following MMS treatment (5-23 micromol/l), FPG induced a dramatic increase in breaks compared with control levels and ENDOIII also showed a significant but smaller increase; in marked contrast, hOGG1 gave no increase. With ENU (0.5-2.0 mmol/l), increases in breaks were seen with FPG and ENDOIII at 1 and 2 mmol/l but, again, no increase was observed with hOGG1. These data indicate that all three endonucleases recognize oxidative DNA damage and, in addition, FPG and ENDOIII also recognize alkylation damage. Therefore, caution should be taken when using FPG and ENDOIII in the comet assay with an agent that has an unknown mode of action since any additional strand breaks induced by either enzyme cannot necessarily be ascribed to oxidative damage. The use of hOGG1 in the modified comet assay offers a useful alternative to FPG and is apparently more specific for 8-oxoguanine and methyl-fapy-guanine.