An expert working group on the in vivo micronucleus assay, formed as part of the International Workshop on Genotoxicity Test Procedures (IWGTP), discussed protocols for the conduct of established and proposed micronucleus assays at a meeting held March 25–26, 1999 in Washington, DC, in conjunction with the annual meeting of the Environmental Mutagen Society. The working group reached consensus on a number issues, including: (1) protocols using repeated dosing in mice and rats; (2) integration of the (rodent erythrocyte) micronucleus assay into general toxicology studies; (3) the possible omission of concurrently‐treated positive control animals from the assay; (4) automation of micronucleus scoring by flow cytometry or image analysis; (5) criteria for regulatory acceptance; (6) detection of aneuploidy induction in the micronucleus assay; and (7) micronucleus assays in tissues (germ cells, other organs, neonatal tissue) other than bone marrow. This report summarizes the discussions and recommendations of this working group. In the classic rodent erythrocyte assay, treatment schedules using repeated dosing of mice or rats, and integration of assays using such schedules into short‐term toxicology studies, were considered acceptable as long as certain study criteria were met. When the micronucleus assay is integrated into ongoing toxicology studies, relatively short‐term repeated‐dose studies should be used preferentially because there is not yet sufficient data to demonstrate that conservative dose selection in longer term studies (longer than 1 month) does not reduce the sensitivity of the assay. Additional validation data are needed to resolve this point. In studies with mice, either bone marrow or blood was considered acceptable as the tissue for assessing micronucleus induction, provided that the absence of spleen function has been verified in the animal strains used. In studies with rats, the principal endpoint should be the frequency of micronucleated immature erythrocytes in bone marrow, although scoring of peripheral blood samples gives important supplementary data about the time course of micronucleus induction. When dose concentration and stability are verified appropriately, concurrent treatment with a positive control agent is not necessary. Control of staining and scoring procedures can be obtained by including appropriate reference samples that have been obtained from a separate experiment. For studies in rats or mice, treatment/sampling regimens should include treatment at intervals of no more than 24 hr (unless the test article has a half‐life of more than 24 hr) with sampling of bone marrow or blood, respectively, within 24 or 40 hr after the last treatment. The use of a DNA specific stain is recommended for the identification of micronuclei, especially for studies in the rat. In the case of a negative assay result with a non‐toxic test article, it is desirable that systemic exposure to the test article is demonstrated. The group concluded that successful application of automated scoring by...