Oral squamous cell carcinoma (OSCC) represents the most frequent of all oral neoplasms in the world. Genetics plays an important role in the etiopathogenesis of OSCC. However, the investigation of the molecular mechanism of OSCC is still incomplete. In this article, we introduced a new approach to detect OSCC-associated genes, in which we not only compare mean difference, but also variance difference between cases and controls. Based on two OSCC datasets from Gene Expression Omnibus, we identified 456 differentially variable (DV) gene probes, in addition to 2,375 differentially expressed (DE) gene probes. There are 2,193 DE-only probes, 274 DV-only probes, and 182 DE-and-DV probes. DAVID functional analysis showed that genes corresponding to DE-only, DV-only, and DE-and-DV probes were enriched in different KEGG pathways, indicating they play different roles in OSCC. This new approach can be used to investigate the genetic risk factors for other complex human diseases.