Background: Soft tissue sarcomas (STSs) are rare, mesenchymal tumors that are primarily treated surgically with radiotherapy (RT) added to improve local control. Selective pressures from the tumor microenvironment drive clonal evolution that can substantially impact the response of cancers to therapy. We integrated multi-region sequencing and circulating tumor DNA (ctDNA) analysis to quantify STS clonal selection in the absence and presence of RT. Methods: We retrospectively identified 11 patients with high-grade pleomorphic sarcomas treated with preoperative RT, including 7 patients with paired pre-RT and post-RT samples, and 9 patients treated with upfront surgery. Multi-region whole exome sequencing of 117 spatially distinct tumor regions and matched germline DNA was performed to identify single nucleotide variants (SNVs) and copy number alterations. Personalized hybrid capture panels were designed to enrich for each patient’s SNVs for ultra-deep targeted sequencing of tumor samples and ctDNA analysis. We inferred the subclonal clusters within untreated (no-RT) and irradiated (post-RT) STSs using PyClone. In addition, we evaluated 6 metrics of intratumoral heterogeneity (ITH) using simulated tumors with a similar mutation rate to human STSs and increasing levels of positive selection. We compared ITH metrics for no-RT human STSs with simulated tumors that developed under neutral evolution and post-RT human STSs using two-sided Mann-Whitney U tests. Results: We observed a similar number of mutations and similar mutational signatures between no-RT and post-RT STSs. Analysis of paired tumors samples identified subclonal expansion and/or contraction after RT in all patients analyzed. We identified 3 ITH metrics that significantly increased in simulated STSs with positive selection, and one ITH metric with a dynamic range sufficient to detect increased selection after RT in STSs (Wright’s fixation index: FST). No-RT STSs had significant evidence of positive clonal selection by all 3 ITH metrics in comparison to simulated tumors under neutral evolution. Furthermore, we observed a significant increase in positive selection after RT (median FST: no-RT 0.08 vs. post-RT 0.17, P=0.04). Conclusions: STSs undergo strong positive selection during tumor development, and RT further increases selection for resistant subclones. Characterizing baseline tumor heterogeneity and developing approaches to sensitize resistant subclones could improve responses to RT.