Identification and verification of clinically actionable molecular variations to refine currently adopted risk-stratified treatment strategy for esophageal squamous cell carcinoma (ESCC) is urgently needed. Here, we evaluated FGFR3 amplification status by fluorescence in situ hybridization (FISH) performed on tissue microarrays and its prognostic value in 526 ESCC patients. FGFR3 amplification was found in 3.0% (16/526) of ESCC patients enrolled in this study cohort. Intratumor heterogeneity and metastatic heterogeneity of FGFR3 amplification were found in 10% (2/20) and 40% (2/5) FGFR3 amplified ESCC cases, respectively. No statistically significant associations were found between FGFR3 amplification status and common clinicopathological features. Survival analyses demonstrated that FGFR3 amplification was associated with a worse disease-free survival (DFS) and overall survival (OS) (DFS, P = 0.008; OS, P = 0.027). Univariate and multivariate analyses revealed that invasive depth was significantly associated with DFS (P = 0.001, HR: 1.498, 95% CI: 1.172-1.914) and OS (P = 0.002, HR: 1.482, 95% CI: 1.159-1.894), and FGFR3 amplification was significantly associated with DFS (P = 0.020, HR: 2.065, 95% CI: 1.120-3.808) and tend to associate with OS (P = 0.070, HR: 1.756, 95% CI: 0.954-3.233). Furthermore, when patients were stratified into stage I-II group and stage III-IV group, the adverse effect of FGFR3 amplification on prognosis was presented in stage III-IV patients (DFS, P = 0.0047; OS, P = 0.029) rather than stage I-II patients (DFS, P = 0.46; OS, P = 0.53), indicating that the prognostic value of FGFR3 amplification may relying on clinical stage. Our findings might provide a better understanding of the FGFR3 amplification status in ESCC patients and add further insights into its potential prognostic value.