Mobile devices have become ubiquitous in our daily lives. They are used for a variety of tasks, ranging from communication to entertainment. In addition, mobile devices are increasingly used for storing valuable personal information. Biometric authentication is a promising approach to increase the security of stored secrets while maintaining a high level of usability. However, biometric data is sensitive and must be protected. Homomorphic encryption is a promising approach to protect biometric data because it allows to perform computations on encrypted data without decrypting it. This thesis is concerned with the extension of a homomorphic encryption scheme for multi-biometrics to support adaptive verification. The implementation was evaluated in terms of biometric performance, computational performance as well as privacy. Our results show that the proposed extension combines the security advantages of storing extracted features concatenated with the flexibility to adapt the verification threshold at runtime in order to to spare the user unnecessary modality presentations for actions that only require a modest level of security. Our contribution therefore improves both usability and computational performance of biometric authentication in the mobile scenario.