Background: Targeted therapies for angiosarcoma (AS) patients have limited efficacy. Although significant responses to immunotherapy (IO-therapy) have been observed in cutaneous AS, its efficacy across all types of AS is not known. Herein, we describe genetic and molecular biomarkers of AS in order to propose potential therapeutic options. Methods: We retrospectively reviewed 143 AS tumors profiled by Next-Generation Sequencing (NGS) 592-gene panel (Caris Life Sciences,Irving, TX, USA). Whole transcriptome sequencing (WTS) was performed on 53 tumors. Mutations and copy number amplifications (CNAs) were analyzed and grouped by pathway. Biomarkers potentially associated with response to IO-therapy (TMB-High [≥10/Mb], MSI-High, and PD-L1 [IHC ≥ 2+ and 5%]) were also analyzed. AS subtypes based on primary tumor site were compared. P-values were corrected using a Benjamini & Hochberg method. Results: Sample median age was 67 (range 22-89), 61% were female, and 29% were classified as metastatic/recurrent. The most commonly mutated genes were TP53 (29%), ARID1A (17%), POT1 (16%), and ATRX (13%); MYC CNA was found in 23% of cases. IO-therapy markers were present in 36.4% of cases (TMB-High in 26%, PD-L1+ 21.8%, MSI-High 0.7%). Pathway alterations were detected in 86% of AS cases. By pathway, TP53 was altered in 31%, cell cycle 30%, DNA damage repair (DDR) 21%, RAS 18%, PI3K 15% and chromatin remodeling 14%. By site, head/neck (HN) AS had the highest rate of IO-therapy markers (65%, p < 0.05) [TMB-High (63%, p < 0.0001)], TP53 mutation (51%, p = 0.07), and POT1A mutation (41%, p < 0.01). MYC CNA was highest in breast (63%) and extremity (40%) AS (p < 0.0001). DDR alterations were present in 56% (p = 0.09) of cutaneous AS and ranged from 12-27% in other subtypes (not significant, NS). RAS and PI3K alterations ranged from 6-27% across all subtypes (NS). Conclusions: Our findings suggest differential angiosarcoma biology across primary sites. HN AS had more frequent markers of potential IO-therapy response, as well as DDR alterations. Next in frequency, we found ARID1A which is possibly associated with overactive EZH2, a target of tazemetostat. MYC amplification suggests a role targeting cell cycle via cyclin-dependent kinase or bromodomain inhibitors in breast and extremity ASs. Finally, RAS and PI3K are mutated in a low percentage of cases, explaining the limited benefit of tyrosine kinase inhibitors in AS. Future AS clinical trials should be designed with consideration of primary site, IO-therapy response biomarkers, and activated pathway.