The aim of the reported study was to investigate the reproducibility of the single-cell gel electrophoresis (SCGE) assay in the determination of DNA single-strand breaks (SSBs) and to estimate the statistical requirements when the SCGE assay is used for the detection of genotoxicity in humans. In human peripheral mononuclear leukocytes (PMLs), we repeatedly measured the rate of SSBs after in vitro incubation of cells for 1 h at 4 degrees C in phosphate buffered saline (PBS, basal) or 10 microM or 50 microM H2O2 (induced). Intra-assay variation was determined from cryopreserved PMLs of a single donor. To assess intrasubject and intersubject variation, PMLs of ten healthy, nonsmoking subjects (aged 19-37 years) were tested 5-9 times. Cryopreserved cells revealed a mean coefficient of variation of 18% (PBS) and 7%-9% (H2O2). There were statistically significant differences between individuals in the rate of SSBs after incubation in PBS (P < 0.01), 10 microM H2O2 (P < 0.001), and 50 microM H2O2 (P < 0.001). The range of interindividual variability was 26% for basal and 12%-13% for induced SSBs, and the coefficient of intraindividual variation was 18%-72% (PBS) and 7%-23% (H2O2). Neither basal nor induced rates of DNA damage were related to gender or age. Estimates of the minimum detectable effects were based on these observed sources of variability (power 90%, level of significance 5%, assumed sample size 50). With two different groups, a difference of 31% in basal SSBs or 12% in induced SSBs would be detectable. Repeated measurement within one group could detect a difference of 26% in basal and 9% in induced SSBs. In summary, the SCGE assay appears to be suitable for the detection of single-strand breaks, e.g., in biomonitoring or environmental medicine, and the statistical requirements could be derived from our analysis of the sources of variability.