The prestress loss estimation provision in the AASHTO LRFD Bridge Design Specifications was recalibrated in 2005 to be more accurate for "high-strength [conventional] concrete." Greater accuracy may imply less conservatism, the result of which may be flexural cracking of beams under service loads. Concern with a potential lack of conservatism and the degree of complexity of these recalibrated prestress loss estimation provisions prompted the investigation to be discussed in this dissertation. The primary objectives of this investigation were: (1) to assess the conservatism and accuracy of the current prestress loss provisions, (2) to identify the benefits and weaknesses of using the AASHTO LRFD 2004 and 2005 prestress loss provisions, and (3) to make recommendations to simplify the current provisions. These objectives were accomplished through (1) the fabrication, conditioning, and testing of 30 field-representative girders, (2) the assembly and analysis of a prestress loss database unmatched in size and diversity when compared with previously assembled databases, and (3) a parametric study investigating the design implications and sensitivity of the current loss provisions. Based on the database evaluation coupled with the experimental results, it was revealed that the use of the AASHTO LRFD 2005 prestress loss provisions resulted in underestimation of the prestress loss in nearly half of all cases. A loss estimation procedure was developed based on the AASHTO LRFD 2005 provisions to greatly simplify the procedure and provide a reasonable level of conservatism. / text