Accurate prediction of prestress losses is a very important step in the design of a highly stressed high-performance concrete (HPC) girder and can affect the service behavior of the girder, such as deflections, camber, and cracking. Current methods for calculating prestress losses according to AASHTO and the Prestressed Concrete Institute were developed for conventional concrete. Further research is needed to determine if the current empirical equations provide an accurate estimate of prestress losses for an HPC girder. Actual prestress losses in HPC girders need to be measured and compared with the predicted losses by use of these equations. The higher the prestressing force in the girder, the larger the concrete compressive strength that is needed at the time of release of the prestressing strands. To help achieve the higher release strength, the precast plants have been using longer curing times. Many precast plants are using steam curing to increase the curing rate. A better understanding of the effects of this heating on HPC is needed. An optical fiber monitoring system was designed and built into a three-span HPC highway bridge. The Rio Puerco Bridge, located 15 miles west of Albuquerque, New Mexico, is the first bridge to be built with HPC in New Mexico. The bridge has three spans each with a length of 29 to 30 m. It is designed to be simply supported for dead load and continuous for live load. HPC was used for the cast-in-place concrete deck and the prestressed concrete beams. A total of 40 long-gauge (2-m-long) deformation sensors, along with thermocouples, were installed in parallel pairs at the top and bottom flanges of the girders. The embedded sensors measured temperature and deformations at the supports, at the quarter spans, and at midspan. Measurements were collected during beam fabrication (casting of the beams, steam curing, strand release, and storage), bridge construction, and servicing. The data collected were analyzed to calculate the prestress losses in the girders, to compare the losses with the predicted losses by available code methods, and to get a better understanding of the properties and behavior of HPC.