We present a valid-time extension of SQL and investigate its efficient implementation on an Object-Relational database system. We propose an approach where temporal queries are expressed using a point-based time model, which only requires minimal extensions to SQL:1999. Our prototype system called TENORS (for Temporal ENhanced Object-Relational System) maps the external point-based temporal queries and data model into equivalent internal representations based on time intervals. We describe the mapping of queries from external views to internal relations, and the temporal clustering and indexing methods used to support these queries on DB2.