Disclosed are a user identity recognition method, a recognition system and a health instrument. The method comprises: extracting more than one physical sign parameters of a user to be recognized (101) combining each physical sign parameter with historical physical sign parameters corresponding to various stored users in a user group respectively and then conducting an analysis to obtain a correlation probability of each physical sign parameter and each stored user (102) according to the preset weight of each physical sign parameter, conducting a comprehensive analysis on the correlation probabilities of all the physical sign parameters and the same stored user, to obtain corresponding probabilities of all the physical sign parameters and the various stored users, respectively (106) and comparing all the corresponding probabilities, and determining the stored user to which the largest corresponding probability corresponds as the identity of the user to be recognized (107). According to the method, a comprehensive analysis is conducted on a plurality of physical sign parameters on the basis that the physical sign parameters of users are stable for a long term and the same physical sign parameters of different users are somewhat different, so that the accurate identity of the user to be recognized can be obtained through fuzz computation.Linvention concerne un procédé de reconnaissance didentité utilisateur, un système de reconnaissance et un instrument de santé. Le procédé comprend les étapes consistant à : extraire plus dun paramètre de signes physiques dun utilisateur à reconnaître (101) combiner chaque paramètre de signes physiques à des paramètres de signes physiques historiques correspondant à différents utilisateurs stockés dans un groupe dutilisateurs, respectivement, puis réaliser une analyse pour obtenir une probabilité de corrélation de chaque paramètre de signes physiques et de chaque utilisateur stocké (102) selon le poids prédéterminé de chaque paramètr
