The two-valued unapproximable trapdoor predicate of probabilistic encryption is extended to the multivalued unapproximable trapdoor predicate in this paper. The corresponding encryption and deoryption algorithm are given. It is proved that the extended pr