Virtually all known tumor predisposing genes have been identified via the analysis of familial cancer cases. Here we argue that this approach is likely to miss recessively acting cancer genes and suggest the analysis of family history-negative patients with multiple primary malignancies for identifying homozygous at-risk genotypes. We performed calculations showing that the homozygous carriers of rare recessive cancer predisposing alleles are unlikely to report a family history of the disease. We further revealed that the c.2515_2519delAAGTT homozygous mutation in a Holliday junction resolvase, GEN1, was overrepresented in women with bilateral breast cancer (BC) as compared to healthy controls [11/360 (3.1 %) vs. 18/1305 (1.4 %); odds ratio (OR) = 2.25 (1.02-4.75); p = 0.031], although this trend was not maintained in unilateral BC patients [23/1851 (1.2 %)]. Noticeably, presence of biallelic c.2515_2519delAAGTT mutation was associated with the absence of BC in mother both in bilateral and unilateral BC cases [7/239 (3.0 %) vs. 0/41 (0 %) and 21/1,558 (1.3 %) vs. 0/215 (0 %), respectively; Mantel-Haenszel p = 0.041]. Thus, this study suggests that identification of dominant and recessive cancer predisposing genes may require distinct study groups.