The following summary represents a consensus of the working group, except where noted. The goal of this working group was to identify the minimal requirements needed to conduct a scientifically valid and practical in vivo chromosomal aberration assay. For easy reference, the items discussed are listed in the order in which they appear in OECD guideline 475. Specific disagreement with the current and/or proposed OECD guideline is presented in the text. Introduction, purpose, scope, relevance, application, and limits of test: This test would not be appropriate in situations where there was sufficient evidence to indicate that the test article or reactive metabolites could not reach the bone marrow. Test substances: Solid and liquid test substances should be dissolved, if possible, in water or isotonic saline. If insoluble in water/saline, the test substance should be dissolved or homogeneously suspended in an appropriate vehicle (e.g., vegetable oil). A suspension was not considered suitable for an intravenous injection. The use of dimethyl sulfoxide as an organic solvent was not recommended. The use of any uncommonly used solvent/vehicle should be justified. Freshly prepared solutions or suspensions of the test substance should be employed unless stability data demonstrate the acceptability of storage. Selection of species: Any commonly used rodent species was deemed acceptable but rats or mice were preferred, with no strain preference. Number and sex: A consensus could not be reached as to the requirement for both sexes versus one sex in this assay. It was suggested that a single sex should be used unless pharmacokinetic and/or toxicity data indicated a difference in metabolism and/or sensitivity between males and females. The size of the experiment (i.e., number of cells per animal, number of animals per treatment group) should be based on statistical considerations. Lacking a formal analysis, it was agreed that at least 100 metaphase cells should be scored per animal while at least five animals of any one sex should be evaluated per treatment group. Recently, a formal analysis of the numbers of cells to score per animal and numbers of animals to score per treatment group was conducted at a workshop on statistics for in vivo mutagenicity tests (Adler et al., 1994). The conclusion of this workshop was that, based on a type I error of 0.05 and a power of 80% to detect at least a doubling in the control frequency, the minimal number of cells to score per animal was 200 and the minimal number of animals to score per sex per treatment group was four. Treatment schedule: Both single and multiple treatment regimens were considered acceptable, as long as a genotoxic effect had been demonstrated or, for a negative study, as long as toxicity had been demonstrated or the limit dose (2000 mg/kg/day for 1-14 days, 1000 mg/kg/day for greater than 14 days) had been used. Dose levels: The dose-finding test should be conducted in the same sex, strain, and species, using the same treatment protocol and solvent/vehicle as proposed for the chromosomal aberration test. The maximum testable dose should be determined independently in males and females. The highest dose tested should be based on mortality, bone marrow cell toxicity, or clinical symptoms of toxicity related to the test compound. If the selection of the highest dose is based on mortality or toxicity, then at least three dose levels spaced by a factor of between 2 and/10 should be used. However, a single 'limit dose' was deemed acceptable if there was no evidence of mortality or toxicity. Controls: A negative control group must be included at all sampling times, while a concurrent positive control group must be included in each test. Performance of the test: When a single treatment protocol is used, bone marrow samples should be collected at two sample times (12-18 h, 36-44 h) after treatment. When a multiple treatment schedule (i.e., two or more administrations) is used, a single sample should be obtained between 12 and 18 h after the last treatment. Analysis: All slides should be coded prior to scoring. The acceptable range of chromosome counts for scorable metaphase cells is 2N ± 1, where N is the haploid number of chromosomes for that species. During scoring, a record should be maintained of the various classes of chromosomal aberrations among cells. The laboratory should state the system of classification being used for scoring aberrations, and provide a clear definition of each type of aberration scored, including gaps. The minimal classes of aberrations to score and categorize would be chromatid-type versus chromosome-type and, among these two categories, gaps, breaks, and rearrangements. Treatment of results: To assess the clastogenicity of the test article, the number of metaphase cells containing at least one chromosomal aberration (including or excluding gaps depending on the criteria used for classifying gaps) should be evaluated. It was agreed that an appropriate statistical analysis combined with scientific judgement and experience was needed to evaluate the experimental results. A consensus could not be reached as to the merit of one statistical method or another. However, it was agreed that any statistical test for the analysis of clastogenic activity should be one-tailed. The historical negative control data should be used to determine if the concurrent control response was acceptable. Evaluation of results: An experimental test response would be classified as positive if the frequency of aberrant metaphase cells was significantly increased in at least one dose group and if the increase was dose-dependent. If either, but not both, of these conditions were met, the experimental results would be classified as equivocal. If neither of these conditions was met, the experimental results would be classified as negative. A confirmatory test would be required whenever different interpretations of the results are likely. © 1994.