Soft tissue sarcoma (STS) diagnostics and prognostics are challenging, particularly in highly malignant and pleomorphic subtypes such as undifferentiated pleomorphic sarcoma (UPS) and leiomyosarcoma (LMS). We applied 32K BAC arrays and gene expression profiling to 18 extremity soft tissue LMS and 31 extremity soft tissue UPS with the aim of identifying molecular subtype signatures and genomic prognostic markers. Both the gains/losses and gene expression signatures revealed striking similarities between UPS and LMS, which were indistinguishable using unsupervised hierarchical cluster analysis and significance analysis for microarrays. Gene expression analysis revealed just nine genes, among them tropomyosin beta, which were differentially expressed. Loss of 4q31 (encompassing the SMAD1 locus), loss of 18q22, and tumor necrosis were identified as independent predictors of metastasis in multivariate stepwise Cox regression analysis. Combined analysis applying loss of 4q31 and 18q22 and the presence of necrosis improved the area under receiver operating characteristic curve for metastasis prediction from 0.64 to 0.86. The extensive genetic similarities between extremity soft tissue UPS and LMS suggest a shared lineage of these STS subtypes and the new and independent genetic prognosticators identified hold promise for refined prognostic determination in high-grade, genetically complex STS.