Current requirements for control of live viral vaccines, including yellow fever 17D, produced from potentially neurotropic wild-type viruses include tests for neurovirulence in nonhuman primates. We have used yellow fever 17D virus as a live vector for novel flavivirus vaccines (designated ChimeriVax) against dengue, Japanese encephalitis (JE), and West Nile (WN) viruses. For control of these vaccines, it would be preferable to substitute a test in mice for the test in a higher species (monkeys). In this study, we compare the neurovirulence of ChimeriVax vaccine candidates in suckling mice inoculated by the intracerebral (IC) route with graded doses of the test article or yellow fever 17D vaccine as a reference control. Mortality ratio and survival distribution are the outcome measures. The monkey safety test is performed as described for control of yellow fever vaccines. In both mice and monkeys, all chimeric vaccines were significantly less neurovirulent than yellow fever 17D vaccine. The test in suckling mice discriminated between strains of two different vaccines (ChimeriVax-JE and ChimeriVax-DEN1) differing by a single amino acid change, and was more sensitive for detecting virulence differences than the test in monkeys. The results indicate that the suckling mouse test is simple to perform, highly sensitive and, with appropriate validation, could complement or possibly even replace the neurovirulence component of the monkey safety test. The test in infant mice is particularly useful as a means of demonstrating biological consistency across seed virus and vaccine lots.