The genus Flavivirus includes a number of important human pathogens that impose major health problems in large regions of the world. The emergence of flaviviruses in new geographic regions (e.g., West Nile virus in North America) and rapid socioeconomic changed in many developing countries where flaviviruses such as dengue virus and Japanese encephalitis virus and endemic demand the development of new vaccines against these diseases. Using tick-borne encephalitis virus as a model we have established a new method to generate attenuated flavivirus strains that may be useful for generating cost-effective and safe live vaccines. This method relies on the specific introduction of deletions into one of the structural proteins, the capsid protein C. These deletions remove parts or all of an internal stretch of hydrophobic amino acid residues that probably is involved in virion assembly. We observed that remarkably long deletions were tolerated, yielding viable viral mutants that were highly attenuated in the mouse model but efficiently induced protective immunity. Biochemical analyses suggested that attenuation was caused by an assembly defect of infectious virions but the mutants produced ample amounts of non-infections subviral particles. The generation of viable mutants with deletions longer that 16 amino acid residues depended on additional, spontaneously emerging mutations within protein C that increased the hydrophobicity of the mutant protein. Although the second-site mutations increased infectivity, they did not restore neuroinvasiveness. Mouse experiments demonstrated excellent safety and immunogenicity profiles for these mutants.