This paper studies Trust in real-time distributed database systems. High availability and fault-tolerance are important aspects of these real-time systems and are commonly achieved by replication. Different existing replication techniques are compared in this paper in terms of implementation and limitations. Also, a new replication technique; TIRS; is proposed that relies on Trust when reading or writing data. Adding Trust to the replication protocol enabled TIRS to perform well in the presence of faulty or malicious nodes. Performance analysis was done using Matlab simulation. TIRS is proved to be fault-tolerant and very fast when reading data. Furthermore, some initial simulations were done to analyze the performance of the write operation in TIRS therefore proving its efficiency.