Abstract. Data integrity is an assurance that data has not been modified in an unknown or unauthorized manner. The goal of this paper is to allow a user to leverage a small amount of trusted client-side computation to achieve guarantees of integrity when interacting with a vulnerable or untrusted database server. To achieve this goal we describe a novel relational hash tree, designed for efficient database processing, and evaluate the performance penalty for integrity guarantees. We show that strong cryptographic guarantees of integrity can be provided in a relational database with modest overhead.