The invention discloses an information recommendation method and system, which are applied between a client and a server. The method comprises a user information collection step: the client divides user information into a plurality of user information groups, and independently stores the user information groups; a request sending step: when a user initiates a recommendation request at the client, the client extracts at least one group in the user information groups for random splicing, and after generating at least one user information combination, the user information combination is encrypted and sent to a server side; and a recommendation result step: after the server side decrypts the user information combinations, randomly distributing each user information combination to a plurality of recommendation servers, and returning a recommendation result to the client based on the user information combinations by the recommendation servers to finish information recommendation to the user. According to the method, dispersed storage of user privacy data is realized, and meanwhile, the recommendation effect of a recommendation system is considered.