High grade prostatic intraepithelial neoplasia (HGPIN) is considered a neoplastic lesion that precedes prostate cancer (PCA). Genetically, HGPIN is known to have some driver mutations such as TMPRSS2-ERG fusion and 8p loss. However, we still do not know how the whole picture of somatic mutations in HGPIN is and what are added during the progression to PCA. To identify the genomic differences between HGPIN and PCA, we analyzed 20 regions of paired HGPIN and PCA from 6 patients by whole-exome sequencing and array-comparative genomic hybridization. As expected, the number of total mutations and CNAs of HGPINs were significantly fewer than those of PCAs. Mutations in FOXA1 and CNAs (1q and 8q gains) were detected in both HGPIN and PCA (common), suggesting their roles in early PCA development. Mutations in SPOP, KDM6A, and KMT2D were PCA-specific, suggesting their roles in HGPIN progression to PCA. The 8p loss was either common or PCA-specific. HRAS and FOXA1 mutations were PIN-specific. Our data show that PCAs are direct descendants of HGPINs in most cases that require more genomic alterations to progress to PCA. The nature of heterogeneous HGPIN population that might attenuate genomic signals should further be studied. Our results provide a clue to explain the long latency from HGPIN to PCA and provide useful information for clinical decision making in the genetic diagnosis of HGPIN and PCA.