The alkaline comet assay, when employed as a genotoxicity test, has relatively low sensitivity because it fails to detect--at non-cytotoxic concentrations--known genotoxins that do not induce breaks or alkali-labile sites. We demonstrate that this limitation is overcome by incorporating in the assay the DNA repair enzyme formamidopyrimidine DNA glycosylase (FPG) to convert damaged bases to breaks. We tested 11 chemicals in human TK-6 cells: three non-cytotoxic--D-mannitol, Tris and EDTA; two cytotoxic--Triton X-100 and fluometuron; and six genotoxic--methylmethanesulphonate (MMS), methylnitrosourea (MNU), cyclophosphamide, benzo(a)pyrene, 4-nitroquinoline-1-oxide (4NQO) and etoposide. At concentrations of MMS, MNU, benzo(a)pyrene or 4NQO causing little or no cytotoxicity and few if any DNA breaks, FPG substantially enhanced the cellular response. Etoposide increased breaks but not FPG-sensitive sites. Cyclophosphamide, a DNA cross linker, gave a response without FPG at 1 μM, but there was no increase with FPG. Triton X-100-induced breaks were secondary to cytotoxicity. The remaining compounds induced no damage. Thus, FPG enhances sensitivity of the comet assay without compromising selectivity.