Comprehensive genetic panel testing generally requires that the analyzed tissues have a percent tumor nuclei (%TN) content of 20% or more to achieve assay performance comparable to the validated specifications. Pathologists play a crucial role in ensuring that the optimal results are achieved by accurately assigning %TN content of the available specimens and selecting the best material to submit for sequencing. This study addresses the issues in evaluating %TN, such as intraobserver variability, and examines whether focused training and feedback can improve pathologist performance. Nine referring institution pathologists (all board-certified and working at the core institute and the alignment hospitals under the National Cancer Genome scheme) evaluated 18 tumors that had been subjected to comprehensive genetic panel testing with the FoundationOne CDx assay. The %TN estimates provided by referring institution pathologists were compared with two standards: %TN assigned by the tumor sequencing institution's pathologist (a board-certified pathologist at Foundation Medicine, Inc.) and the computational %TN estimated from the mutant allele frequencies after sequencing was completed. The pathologists generally overestimated %TN in the first pretraining round of the evaluation, and the differences in the averaged %TN from the tumor sequencing institution and computational standards were statistically significant. However, the posttraining second-round results became significantly concordant with the standards. This study suggests that %TN content is empirically overestimated but the evaluation skill can be improved by providing a training and feedback program.