There is serious concern about the potential harmful effects of certain nanomaterials (NMs), on account of their ability to penetrate cell membranes and the increased reactivity that results from their increased surface area compared with bulk chemicals. To assess the safety of NMs, reliable tests are needed. We have investigated the possible genotoxicity of four representative NMs, derived from titanium dioxide, zinc oxide, cerium oxide and silver, in two human cell lines, A549 alveolar epithelial cells and lymphoblastoid TK6 cells. A high-throughput version of the comet assay was used to measure DNA strand beaks (SBs) as well as oxidised purines (converted to breaks with the enzyme formamidopyrimidine DNA glycosylase). In parallel, cytotoxicity was measured with the alamarBlue® assay, and the ability of NM-treated cells to survive was assessed by their colony-forming efficiency. TiO and CeO NMs were only slightly cytotoxic by the alamarBlue® test, and had no long-term effect on colony-forming efficiency. However, both induced DNA damage at non-cytotoxic concentrations; the damage decreased from 3 to 24-h exposure, except in the case of CeO -treated A549 cells. ZnO and Ag NMs affected cell survival, and induced high levels of DNA damage at cytotoxic concentrations. At lower concentrations, there was significant damage, which tended to persist over 24 h. The implication is that all four reference metal NMs tested-whether cytotoxic or not-are genotoxic. A full assessment of NM toxicity should include tests on different cell types, different times of incubation and a wide range of (especially non-cytotoxic) concentrations; a test for cell viability should be performed in parallel. Inclusion of Fpg in the comet assay allows detection of indirect genotoxic effects via oxidative stress.