Several models of evolution from primary cancers to metastases have been proposed; but the most widely accepted is the clonal evolution model proposed for colorectal cancer in which tumors develop by a process of linear clonal evolution driven by the accumulation of somatic genetic alterations. Various other models of cancer progression and metastasis have been proposed, including parallel evolution and the same gene model. The aim of this study was to investigate the evolution of metastases from primary cancer in 22 patients diagnosed with high-grade serous epithelial ovarian cancer. We established somatic genetic profiles based on the pattern of loss of heterozygosity, in several different regions of tumor tissue within the primary tumor and metastatic deposits from each case. Maximum parsimony tree analysis was used to examine the evolutionary relationship between the primary and metastatic samples for each patient. In addition, we investigated the extent of genetic heterogeneity within and between metastatic tumors compared with primary ovarian tumors. Our data suggest that most, if not all, metastases are clonally related to the primary tumors. However, the data oppose a single model of linear-clonal evolution whereby a late stage clone within the primary tumor acquires additional genetic changes that enable metastatic progression. Instead, the data support a model in which primary ovarian cancers have a common clonal origin, but become polyclonal, with different clones at both early and late stages of genetic divergence acquiring the ability to progress to metastasis.