Uveal melanoma (UVM) is an adult intraocular malignancy which is the most frequent and has a high tendency for metastasis. This study aims to develop significant differential gene subnetwork between primary and metastatic UVM to identify potential prognostic biomarkers. Differentially expressed genes (DEGs) among three chip datasets were downloaded from Gene Expression Omnibus and identified according to standardization annotation information. Genetic enrichment analyses were utilized to describe biologic functions. The protein‐protein interaction network of DEGs was developed and the module analysis was constructed by STRING and Cytoscape. Kaplan‐Meier method of the integrated expression score was applied to analyze survival outcomes. Functional annotation was assessed to perform GO and Kyoto Encyclopedia of Genes and Genomes enrichment analysis. In addition, ClueGO and gene set enrichment analysis were analyzed to detect underlying significant genes and involved signaling pathways. A total of 103 DEGs with function enrichment were recognized and might be considered as candidate prognostic biomarkers between primary and metastatic UVM. Furthermore, Kaplan‐Meier method suggested that SCD5, SPTBN1, FABP5, SQLE, PTPLA (HACD1), and CDC25B were independent prognostic factors in UVM. Functional annotations indicated that the most involved significant pathways including interferon‐gamma response, IL‐6 JAK STAT3 signaling, TNFA signaling via NFKB and inflammatory response. Significant DEGs between primary and metastatic UVM tissue were identified and might have involved in the metastasis of UVM. SCD5, SPTBN1, FABP5, SQLE, PTPLA (HACD1), and CDC25B transcription levels were of high prognostic value, which might assist us to understand the underlying carcinogenesis or advancement of UVM better.