Data replication is often used in distributed database applications to improve data availability and performance. Replicated data must be periodicall- y refreshed using update propagation strategies. Most of the current strategie- s guarantee mutual consistency of replicated data but are inefficient. Lazy (or asynchronous) replication is an alternative, more efficient solution where mutual consistency is relaxed. It is needed in applications such as on-line financial transactions and telecommunication systems which require high freshness of replicated data. In this case, the concept of \it freshness