In view of shortage of the existing fingerprintbased key hiding method, a new key hiding scheme based on fingerprint is proposed, which fully uses the information of minutiae set and the texture information around every minutia, depending on a new fingerprint model: minutiae texture strings model. The scheme is called minutiae texture strings key hiding scheme. In this scheme, minutiae set is extracted from fingerprint firstly, and Gabor filter is then used to the area around every minutia to extract texture information. Minutiae set and texture information around every minutia construct the minutiae texture strings model. Secret key generated by a symmetric encryption system or distributed by PKI is divided into n shares by a (n, k) secret sharing algorithm, which are then hidden by minutiae texture strings secretly. Query fingerprint can only recover the hidden key when at least k shares of key are retrieved. Experiments are made on FVC2002 DB1 and DB2, in which one fingerprint is used to hide key and another fingerprint is used to recover the key. Equal error rate (EER) of the scheme is no more than 1%~2.2%, which is better than that of normal fuzzy vault. Security analysis of the scheme shows that information of the key and the fingerprint model is protected effectively, with security higher than that of normal fuzzy vault schemes.