Background: Recent studies have confirmed that immune-associated genes perform a crucial function in recurrence and metastasis of thyroid carcinoma. A reliable immune-related prognostic signature for patients with thyroid cancer is needed. This study constructed a novel immune-related prognostic signature for thyroid cancer and evaluated its prognostic value by bioinformatics analysis. Methods: In this study, we anatomized differentially expressed immune-associated genes from The Cancer Genome Atlas database. The samples from The Cancer Genome Atlas database were randomly divided into training set and test set. A novel immune-related prognostic signature for thyroid cancer was developed by least absolute shrinkage and selection operator and Cox regression analysis: Risk score = (0.6846 × expression value of C-X-C motif chemokine ligand 5 [CXCL5]) + (1.1556 × expression value of Azurocidin 1 [AZU1]) + (−0.3156 × expression value of nucleotide binding oligomerization domain containing 1 [NOD1] + (0.0542 × expression value of TNF Receptor Superfamily Member 11b [TNFRSF11B]) + (0.0952 × expression value of VGF nerve growth factor inducible [VGF]). The established prognostic signature was evaluated based on training set and test set by survival curves, receiver–operator characteristic curves, risk score, survival status, heatmap, and independent prognostic analysis. Meanwhile, we appraised the correlation between target immune-associated genes and clinical stage, tumor-infiltrating immune cells respectively. Results: Five immune-associated genes were used for constructing an immune-related prognostic signature by least absolute shrinkage and selection operator, univariate, and multivariate analysis. Survival curves, receiver–operator characteristic curves, and independent prognostic analysis showed the signature had significant prediction value. Clinical and immune cell correlation analyses indicated that target immune-associated genes may participate in tumor immune infiltration and tumor progression. Conclusions: We constructed a novel 5 immune-associated genes signature for predicting the prognosis of patients with thyroid cancer, which may help clinical workers evaluate individualized therapy and prognosis.