Background: The intersection between germline and somatic genomics is an evolving field in which germline mutations may predispose to unique patterns of subsequent somatic mutations in cancer. Germline mutations in CHEK2, involved in cell cycle regulation and DNA damage response, are associated with an increased risk of prostate cancer (PCa), while somatic-only CHEK2 alterations in PCa are rare. The association of germline CHEK2 (gCHEK2)-altered PCa with somatic mutations is unknown, and may inform hypotheses about the etiology of these cancers. Methods: Germline DNA sequencing of 1,042 consecutive PCa patients (pts) from the public SignalDB database (www.signaldb.org) was analyzed for prevalence of pathogenic gCHEK2 mutations and was compared to individuals from the general population estimated by the ExAC database (containing 53,105 germline exomes). A separate cohort of 33 PCa pts from Johns Hopkins (JH) with known gCHEK2 mutations and available somatic tumor DNA sequencing (from primary prostatic tissue) was used to assess the association of gCHEK2 mutations with somatic mutations in genes that are recurrently altered in PCa (TP53, RB1, PTEN, ATM, BRCA1/2, and CDK12); the prevalence of these somatic alterations was compared to those in 333 unselected PCa pts from the TCGA cohort. Somatic biallelic inactivation of CHEK2 was analyzed in a subset of pts. After uncovering a potential link between gCHEK2 and somatic CDK12 mutations, we studied a cohort of 69 pts with somatic CDK12 mutations where germline data were also available. Results: 28 of 1,042 (2.7%) PCa pts from SignalDB had a pathogenic gCHEK2 mutation, compared to a population prevalence (in ExAC) of 1.4% (750 of 53,105) (RR 1.9, 95%CI 1.3–2.8, P< 0.001). Strikingly, only 23.8% of pts from SignalDB with gCHEK2 mutations had biallelic inactivation in the tumor. Furthermore, none of the 33 gCHEK2 pts from the JH cohort had evidence of somatic LOH. There were no differences in mutation prevalences involving TP53, RB1, PTEN, ATM, and BRCA1/2 between gCHEK2-altered and non-altered PCa pts. Unexpectedly, 5 of 33 (15%) gCHEK2-altered pts from the JH cohort had a somatic CDK12 mutation, compared to only 3 of 333 CDK12 mutations (1%) in unselected PCa pts from the TCGA cohort (RR 16.8, 95%CI 4.2–67, P< 0.001). Conversely, 11 of 69 (16%) pts with a somatic CDK12 mutation harbored a pathogenic gCHEK2 mutation, compared to 28 of 1,042 (2.7%) unselected PCa pts from SignalDB (RR 5.9, 95%CI 3.1–11.4, P< 0.001). Conclusions: Prostate cancers from gCHEK2-altered pts are infrequently characterized by biallelic CHEK2 inactivation and may be enriched for somatic CDK12 mutations, suggesting a unique mechanism of carcinogenesis that is different from gBRCA2-altered pts. Conversely, somatic CDK12-mutated cancers may be enriched for gCHEK2 mutations. The co-occurrence of CHEK2 and CDK12 mutations suggests a synergistic role in promoting cancer growth.