Transgenic rodent gene-mutation models provide relatively quick and statistically reliable assays for gene mutations in the DNA from any tissue. This report summarizes those issues that have been agreed upon at a previous IWGT meeting [Environ. Mol. Mutagen. 35 (2000) 253], and discusses in depth those issues for which no consensus was reached before. It was previously agreed that for regulatory applications, assays should be based upon neutral genes, be generally available in several laboratories, and be readily transferable. For phage-based assays, five to ten animals per group should be analyzed, assuming a spontaneous mutant frequency (MF) of approximately 3x10(-5) mutants/locus and 125,000-300,000 plaque or colony forming units (pfu or cfu) per tissue per animal. A full set of data should be generated for a vehicle control and two dose groups. Concurrent positive control animals are only necessary during validation, but positive control DNA must be included in each plating. Tissues should be processed and analyzed in a blocked design, where samples from negative control, positive control and each treatment group are processed together. The total number of pfus or cfus and the MF for each tissue and animal are reported. Statistical tests should consider the animal as the experimental unit. Nonparametric statistical tests are recommended. A positive result is a statistically significant dose-response and/or statistically significant increase in any dose group compared to concurrent negative controls using an appropriate statistical model. A negative result is a statistically non-significant change, with all mean MFs within two standard deviations of the control. During the current workshop, a general protocol was agreed in which animals are treated daily for 28 consecutive days and tissues sampled 3 days after the final treatment. This recommendation could be modified by reducing or increasing the number of treatments or the length of the treatment period, when scientifically justified. Normally male animals alone are sufficient and normally at least one rapidly proliferating and one slowly proliferating tissue should be sampled. Although, as agreed previously, sequencing data are not normally required, they might provide useful additional information in specific circumstances, mainly to identify and correct for clonal expansion and in some cases to determine a mechanism associated with a positive response.