When the mouse lymphoma Tk assay (MLA) provides a positive result, its cause can be roughly estimated by examining colony sizes. An increase in the number of large colonies means that the compound tested has point mutational potential, while an increase in small colonies indicates the potential for chromosome aberration. However, it was found to be difficult to clearly judge this in the case of caffeine known as a clastogen lacking the potential of point mutation. In our study, caffeine significantly increased the thymidine kinase (Tk) mutation frequencies derived from large colonies as well as those from small colonies in the standard protocol, although the frequencies derived from a small colony were higher than those from large colonies at higher doses. Therefore, we prolonged the expression period from 2 days, a standard period, to 6 days after treatment and then examined the Tk and Hprt mutations simultaneously. The result showed that caffeine gave a completely negative result on a mutation test for both Tk and Hprt. On the other hand, ethyl methanesulfonate (EMS), a genotoxic carcinogen, showed a positive result for both. Moreover, caffeine and EMS significantly increased the frequencies of micronucleated cells. In conclusion, when MLA gives a positive result and the cause is ambiguous, in order to identify the exact cause of the positive response, it is helpful to perform a confirmatory test investigating the potential of Tk and Hprt gene mutation simultaneously after 6-day expression and to perform an in vitro micronucleus assay during the expression period.