Uveal melanoma (UM) has emerged as one of the most common primary intraocular malignant tumors worldwide. Long non-coding RNAs (lncRNAs) are increasingly recognized as decisive factors in the progression and metastasis of UM, involving in epithelial-mesenchymal transition (EMT) of UM. We conducted a comprehensive analysis of lncRNAs closely associated with EMT-related genes in the TCGA UM cohort, identifying 961 EMT-related lncRNAs. Through univariate COX analysis, we identified 9 survival-related EMT-related lncRNAs (sER-lncRNAs), further establishing an EMT-related risk scoring model (ER-RSM) with two sER-lncRNAs (ZNF667-AS1 and AP005121.1) identified by multivariate COX analysis. Through this ER-RSM, low-risk UM patients achieved better overall survival than high-risk UM patients. AP005121.1 was positively correlated with higher stage and M staging in UM patients, while ZNF667-AS1 was positively correlated with earlier stage, T, and M staging in UM patients. , AP005121.1 expression was higher in UM tumor tissues and cell lines than in adjacent normal tissues and human retinal pigment epithelial cells, whereas ZNF667-AS1 expression showed the opposite pattern. siR-AP005121.1 significantly inhibited migration and invasion ability of UM cells and suppressed the EMT pathway, while siR-ZNF667-AS1 promoted migration and invasion of UM cells and activated the EMT pathway. In this study, we screened sER-lncRNAs and constructed an ER-RSM to investigate the relationship between sER-lncRNAs and prognosis and clinical staging of UM. Additionally, we validated the expression of sER-lncRNAs in UM clinical samples and cell lines. The ER-RSM may provide potential key insights for the diagnosis and therapeutic intervention of UM patients.