Different fracture-based tests and parameters have been developed in recent years to assess thermal cracking performance of asphalt mixtures. Agencies have adopted some of these tests, established threshold values, and incorporated them into their specifications to evaluate mixes during design and production. Because of the heterogeneous nature of asphalt mixtures, there can be a substantial amount of variability associated with mechanical test results obtained from fracture-based tests. To obtain unbiased properties with variabilities only inherent to the material itself and not caused by mixture production, specimen fabrication, or testing related factors, an appropriate number of replicate specimens need to be tested. This article provides the results of a study to establish a practical number of replicate tests that are required to obtain accurate and representative asphalt mixture fracture energy from the disk-shaped compact tension (DCT) fracture test for routine use by producers and agencies. Statistical analyses were performed to determine the precision and accuracy associated with the use of 4, 8, and 12 replicates. A total of 368 fracture energy measurements (16 replicates for 23 mixtures) were used for the analysis. The statistical analysis consistently indicated that 12 replicates results in higher confidence in the estimate of the mean. The maximum difference between measurements when 4 replicates are used ranges from 3 to 25 % but reduces to 1 to 9 % for 12 replicates. For 4 replicates, the data set used in this study indicated that 41 % of the time, it was not possible to confidently distinguish whether mixtures met a minimum threshold value or not. However, this error margin drops to 12 % when using 12 replicates. Based on these results, 12-replicate specimens are recommended to ensure necessary accuracy and repeatability for performance-based specifications using DCT fracture energy.