<p><i>Elymus sibiricus</i> L., a key perennial forage grass widely distributed in temperate and alpine regions, holds particular importance on the Qinghai–Tibet Plateau for grassland restoration and livestock production. Efficient assessment of its genetic diversity is therefore fundamental for germplasm conservation and breeding. In this study, we systematically evaluated and compared the efficacy of three functional gene-based marker systems—peroxidase gene polymorphism (POGP), lipoxygenase (LOX), and cytochrome P450 (Cyt P450)—in analyzing the genetic diversity and population structure of <i>E. sibiricus</i>. Using six POGP, eight LOX, and five Cyt P450 primer pairs, we genotyped 54 accessions originating from four major geographical regions. Key genetic diversity parameters—including the total number of bands (TNB), number and percentage of polymorphic bands (NPB and PPB), polymorphism information content (PIC), Shannon diversity index (H), resolving power (RP), marker index (MI), and average band informativeness (Ibav)—were calculated and compared across the markers. Among them, the LOX marker system demonstrated superior performance in most parameters, indicating its higher efficiency in detecting informative genetic variation. Cluster analysis, principal coordinates analysis (PCoA), and Bayesian population structure analysis consistently revealed clear genetic differentiation between accessions from the Qinghai–Tibet Plateau geographic subgroup and those from the other geographic subgroups. This geographically associated genetic structure was further supported by a strong correlation among genetic distance matrices as shown by Mantel tests. Our findings establish that functional gene-based markers, particularly LOX, are highly effective for delineating geographically structured genetic diversity in <i>E. sibiricus</i>, providing valuable tools and insights for germplasm conservation, evolutionary studies, and breeding program design.</p>