This work presents a general secure biometric authentication algorithm that guarantees privacy protection of biometric templates. These are all based on the use of a one-way transform and helper data, which is used to achieve noise robustness at the input of the one-way transform. The biometrics templates obtained during enrollment are described by sequences of n i.i.d. random variables. A hashed form of the template is stored together with helper data in a database that is vulnerable to attacks from the inside.