In this study, a large-scale direct tension specimen (minimum cross-sectional area of 11,000 mm2) was developed to determine its ability to quantify tensile behaviors of fiber-reinforced ultra-high-performance concrete. Direct tension specimens were successfully mixed and molded using readily available equipment in a typical construction materials laboratory. Results from large-scale direct tension tests were compared with results from commonly used indirect methods (i.e., splitting cylinder strength and double punch testing). At early ages, direct tensile specimens had a 0 % failure rate and recorded coefficient of variation values of 6.2 %. Computerized tomography scans from several sections within a large-scale specimen showed sufficiently random fiber orientation, particularly within the desired fracture region. Altogether, this effort showed that large-scale direct tension specimens were successfully able to quantify tensile strength of fiber-reinforced concrete; however, the effects of fiber orientation and boundary conditions on postcracking tensile strength of large-scale specimens need to be further evaluated in future efforts.