Objective. To explore a novel Immune-associated gene signature for overall survival (OS) in patients with oral squamous cell carcinoma (OSCC). Methods. Expression profiles of genes and corresponding clinical materials of OSCC patients were obtained through the TCGA database. With a LASSO Cox regression model, a multigene signature was established to predict the OS of OSCC patients. Some molecular experiments including RNA interference, MTT, and Transwell assay were applied to verify the role of the risk gene FGF9 in OSCC. Results. 43 immune-related prognostic DEGs were identified in OSCC. A 17-gene signature was established to assign the patients to either a high-risk group (HG) or a low-risk group (LG). The HG presented a shorter OS than the LG (
P
<
0.05
). According to multivariate Cox regression analyses, the risk score was considered an independent factor for OS prediction (training set: HR = 3.485, 95% CI = 2.037–5.961,
P
<
0.001
; test set: HR = 4.531, 95% CI = 2.120–9.682,
P
<
0.001
). ROC curve-based analysis revealed the signature’s ability for prediction. According to functional analysis, the immune cell expression and immune function of the HG were significantly inhibited. After knocking down the high-risk gene FGF9, the migration, proliferation, and invasion capabilities of OSCC cells HSC6 were significantly suppressed (
P
<
0.05
). Conclusion. A novel immune-associated gene signature was identified for predicting the prognosis of OSCC. These risk genes show great potential as targets for OSCC treatment.