Currently, it can be argued that in certain areas of information technology, there is a complete replacement of classical computer system user authentication systems based on passwords and tokens with biometric technologies. However, biometric systems are vulnerable to various types of security threats. For example, in them, unlike the same passwords and tokens, templates based on biometrics cannot be replaced in case of compromise. To solve this problem, new protection schemes have been developed. Conventionally, they can be divided into two groups: biometric cryptography and cancelable biometrics. Biometric cryptography methods show average values of errors of the first and second types; experimental work in this area is widely known. Cancelable biometrics can be highly reliable, but there is not much experimental data on them. This paper presents a comparative analysis of the reliability of existing methods. It is shown that among the static biometric parameters the greatest interest is the iris, and among the dynamic ones – the keyboard stroke. However, using these methods, like others, has its own difficulties and risks.