Inconsistent reports of associations between human leukocyte antigen (HLA)-DR and thyroid cancers exist. We conducted a comprehensive search of the PubMed, Scopus and Web of Science databases. Using random-effects modeling, subgroup analyses, meta-regression and prediction interval (PI) estimation, we combined the existing evidence from 13 studies (977 cases of thyroid cancer and 3735 controls). Only HLA-DR1 and HLA-DR11 were significantly associated; however, the evidence for HLA-DR11 came from only three studies while that for HLA-DR1 had large between-study heterogeneity. All the PIs estimated in the study straddled unity. Therefore, current evidence for the studied association is incomplete as well as uncertain. Attempts to include HLA-DR typing as a prognostic or therapeutic marker may be premature at this time.