In a large multilaboratory cytogenetic study of interlaboratory variations using 4 dose levels of triethylenemelamine and a control, a severely damaged cell was operationally defined as a cell which contained at least 10 aberrations of any type. A review of this study suggested that the use of this definition introduced a bias in the measurement and interpretation of results for the other cytogenetic categories studied. As a result, the original severely damaged cells were carefully reanalyzed to investigate the characteristics of this bias and to seek procedures to minimize or eliminate it. Results characterize this bias and demonstrate that when a severely damaged cell is defined as one containing at least 20 aberrations and those aberrations in the remaining non-severely damaged cells are classified by specific type, the bias is significantly reduced and chromosome analysis can be improved as a test system.