PeMS 4 is the latest of four task orders devoted to research, development, and maintenance of the PeMS system. PeMS collects, processes, stores, and makes available online data from the six Caltrans districts (D3, 4, 7, 8, 11, 12), which include the major urban areas in California. The data are obtained from 23,237 loops, grouped into 7,359 vehicle detector stations (VDS). These loops cover 2,812 out of 30,726 miles of interstate and state highways in California. PeMS began as a research project. As the research system evolved, Caltrans determined that the information it provided was very valuable, and significant resources were then directed towards the development of PeMS. This entailed production-level code development and system maintenance. Faculty, post-doctoral fellows, and graduate student researchers at U.C. Berkeley (UCB) conduct the 'research' element of the project. Berkeley Transportation Systems (BTS) is responsible for PeMS 'development' and 'maintenance' activities. The UCB and BTS groups meet weekly. There are periodic conferences with members of Caltrans Division of Traffic Operations. The project is comanaged by the PeMS PI and the Division of Traffic Operations. The product of the research activity is reported in professional meetings and journals and in algorithms, and summarized in this report. The algorithms are incorporated into PeMS applications code as part of the development activity. The maintenance activity is concerned with database management, network troubleshooting, and providing PeMS access to users. PeMS is accessed over a standard web browser via the Internet. The current PeMS user community comprises 956 individuals and 63 'value added resellers' or VARs. During October 2003, PeMS received an average of 4,000 'hits', generating 42 MB of data download, each day. Caltrans users account for 26 percent of hits, and UCB accounts for 12 percent. The remaining usage is by VARs, academic researchers around the country, and transportation professionals. Under PeMS 4, the system changed significantly in three ways. First, the 'look and feel' that users experience has improved. Website navigation is more intuitive. Second, PeMS functionality increased with the incorporation of congestion reports, travel time reliability, loop detector diagnosis, and data correction. Third, the database structure has been changed from a district-by-district orientation to a statewide orientation. As a result, statistics like travel times can be collected for routes that cross district boundaries.