The cytokinesis-block micronucleus assay (CBMN assay) with cultured human lymphocytes is a well-established assay in genotoxicity testing and human biomonitoring. For both approaches, human lymphocytes are stimulated by phytohaemagglutinin (PHA) and cultured for about 72 h; 44 h after PHA stimulation, cytochalasin B (CytB) is added and micronuclei (MN) are scored in binucleated cells. The main difference between these two applications is the way lymphocytes are exposed to mutagens. In order to maximise the probability of detecting a mutagen, the OECD guideline 487 recommends starting the exposure to the test substance at 44-48 h after PHA stimulation. In human biomonitoring, blood samples are taken from subjects exposed to environmental mutagens in vivo and lymphocytes with induced DNA damage at the start of the cell culture are investigated with regard to potentially increased MN frequencies in binuclear lymphocytes. We compared the sensitivity of these two protocols by either treating lymphocyte cultures for 2h with known DNA-damaging mutagens at the start of the culture or 42 h after PHA stimulation. The mutagens used were methyl methanesulfonate (MMS), ethyl methanesulfonate (EMS), N-nitroso-N-ethylurea (ethyl nitrosourea; ENU), styrene oxide (SO), (±)-anti-B[a]P-7,8-dihydrodiol-9,10-epoxide (BPDE) and mitomycin C (MMC). All substances induced MN under the conditions of the standard in vitro CBMN assay but only MMC clearly induced MN in lymphocytes exposed at the start of the culture. All mutagens (except MMC, a known crosslinker) were tested by the comet assay with blood cultures exposed at the start of the culture and clearly induced DNA migration. The nuclear division index (NDI) indicated that damaged lymphocytes proliferated well in these experiments. The lack of increased MN frequencies despite increased damage levels and good proliferation suggests that the CBMN assay is rather insensitive for the detection of mutagens/clastogens when damage is induced at the start of the blood cultures. Potential consequences for the interpretation of human biomonitoring studies are discussed in this article.