Human herpesvirus (HHV)-6 and HHV-7 have been detected in central nervous system and glioma tissue, while their exact role in glioma remains uncertain. Omics profiles and clinical information were downloaded from public databases, including The Cancer Genome Atlas cohort for training set and the Chinese Glioma Genome Atlas cohorts for validation sets. Differentially expressed genes between HHV-6 and HHV-7 infected or noninfected glioma patients were screened for establishing the HHV-6 and HHV-7 infection (HI) model through Lasso regression analysis. Bioinformatics methods were used to analyze the correlation between HI scores and prognosis, metastasis in glioma patients. Predictable efficacy of HI in temozolomide-resistance and HI-related genetic signatures were also explored. The HI model was constructed as: Risk score = (0.014709*DIRAS3) + (0.029787*TEX26) + (0.223492*FBXO39) + (0.074951*MYBL1) + (0.060202*HILS1). The five gene signature showed good performance in predicting survival time for glioma patients, while higher HI score is correlated with malignant features. Moreover, DNA mismatch repair genes were augmented in glioma patients with higher HI score as well as nonresponse to temozolomide treatment, which was in parallel with the transcriptomic result of temozolomide-resistant glioma cell. Targeting the five gene signature is beneficial for prognosis of glioma patients, especially in glioma patients underwent temozolomide treatment.