The California Institute of Technology (Caltech), the United States Geological Survey (USGS), and the California Department of Conservation, Division of Mines and Geology (CDMG) are completing the implementation of TriNet, a modern seismic information system for southern California. TriNet consists of two elements, the Caltech-USGS element and the CDMG element (Mori et al., 1998). The Caltech-USGS element (Caltech-USGS TriNet) concentrates on rapid notification and archiving of data for seismological applications, while the CDMG element is focused on the needs of engineering users (Hauksson et al., 2002). All three. TriNet agencies are working toward facilitating emergency response and long-term mitigation of earthquake hazards in cooperation with other agencies. The technical development of Caltech-USGS TriNet is sufficiently different from the CDMG element of TriNet to warrant a separate description. This paper provides a technical overview of the design principles of Caltech-USGS TriNet. These principles were based on a document that stated the scientific requirements of TriNet (Jones et al., 1997). We also describe the implementation of these principles using modern technology. The implementation consisted of station deployments, establishing communications links, and developing and implementing new hardware and software for data processing and information distribution. Thus, the Caltech-USGS TriNet is an integrated project extending across many disciplines, using basic ground-motion data and seismological algorithms to generate in near real-time a sophisticated earthquake knowledge base following earthquakes in southern California. Caltech-USGS TriNet applies advanced technology to record both small and large earthquakes on scale. The latest generation of broadband and strong-motion sensors with 24-bit digitizers is used to acquire high-fidelity ground-motion data. Real-time communication is a requirement to facilitate rapid processing and notification about seismicity for emergency management. The data acquisition systems are designed to ensure redundancy and automated processing of data. To accomplish automation, high-speed computers and advanced software form the inner workings of the Caltech-USGS TriNet system. Adopting the commercial database Oracle is an important foundation of our data management system. The automated flow of data into an accessible data center and the automatic population of the database is part of our new seismic network design and is an essential feature of Caltech-USGS TriNet. The TriNet real-time systems and database have been operating online for more than two years, processing real-time data currently from more than 375 stations, or more than 1,200 high sample-rate data channels. Many of these capabilities were tested in the 1999 M_w 7.1 Hector Mine earthquake. New postprocessing and catalog-generation approaches have also been implemented in 2001. Caltech-USGS TriNet is one of the first U.S. regional seismic networks that uses digital technology on a scale of 200 or more stations, with both broadband and strongmotion sensors. In comparison, the IRIS Global Seismic Network consists of 108 stations, with plans for a total of 150 stations (Hutt and Bolton, 1999). Previous digital networks, such as TERRAscope (Kanamori et al., 1997) and the Berkeley Digital Seismic Network (BDSN) (Gee et aL, 1996), have been smaller than TriNet, with about 20 stations each. TriNet also benefits from the experience of other seismic networks around the world. The K-Net in Japan is another example of large-scale deployment of a digital network, although it is focused on strong motions (Kinoshita, 1998). Extensive developments of strong-motion networks in Taiwan and associated near-real-time processing of data employ somewhat different technology but have similar goals for information products following large earthquakes (Teng et al., 1997).