Recently, it has been shown by various studies that tumor mutation burden (TMB) can correlate with a cancer patient’s responsiveness to checkpoint inhibitor immunotherapy. These findings have stimulated a widespread interest in the development of cost-effective assays that accurately measure TMB. Here we assess whether TMB measured through next-generation sequencing (NGS) of a 2-megabase 500-gene panel correlates with that determined through whole exome sequencing (WES). We further investigate whether sequencing tumor samples alone combined with a computational germline filtering and background noise removal algorithm can measure TMB accurately compared with sequencing of tumor-normal pairs. TMB was calculated as mutations (single nucleotide variants, insertions, and deletions) per megabase. Using downloaded TCGA data from 2,385 subjects across colon, lung and melanoma cancer types, TMB was measured using tumor-normal WES or the in-silico-filtered tumor-only 500-gene panel and the two approaches correlated linearly (R2 = 0.96). Collectively, our data demonstrates that TMB determined from NGS using the 500 gene panel correlates strongly with TMB measured through WES, with a sensitivity of >90% and a specificity of >90%. In addition, sequencing tumor samples alone combined with bioinformatic germline-filtering and background noise-removal was similar to sequencing subject-matched normal DNA. These findings support the use of a 500 gene panel for accurate measurement of TMB while the addition of bioinformatics enables tumor-only sequencing.