Metastasis is a major factor associated with poor prognosis in cancer, but little is known of its molecular mechanisms. Although the clinical behavior of soft tissue sarcomas is highly variable, few reliable determinants of outcome have been identified. New markers that predict clinical outcome, in particular the ability of primary tumors to develop metastatic tumors, are urgently needed. Here, we have chosen leiomyosarcoma as a model for examining the relationship between gene expression profile and the development of metastasis in soft tissue sarcomas. Using cDNA microarray, we have identified a gene expression signature associated with metastasis in sarcoma that allowed prediction of the future development of metastases of primary tumors (Kaplan-Meier analysis P = 0.001). Our finding may aid the tailoring of therapy for individual sarcoma patients, where the aggressiveness of treatment is affected by the predicted outcome of disease.