Lynch syndrome is a hereditary predisposition to colorectal cancer (CRC) and other cancers caused by germline mutations in mismatch repair (MMR) genes. Almost all the cancers diagnosed in Lynch syndrome have an acquired MMR deficiency, a replication error (RER) mutator phenotype that is also found in a minority of sporadic cancers developed in the target organs of Lynch syndrome. Lynch syndrome displays many curious features that cannot be accounted for by the prevailing concepts of carcinogenesis and genetics: (1) CRCs occur preferentially in the right side of the colon, whereas the majority of sporadic cases develop in the left colon; (2) the increased risk of CRC is not associated with an increased incidence of adenomatous polyps, which are necessary precancerous lesions in the development of common CRCs; (3) the tumor spectrum in Lynch syndrome is restricted to the colon and some extracolonic sites, whereas the responsible MMR genes are ubiquitously expressed; (4) the tumor risk, which is negligible during childhood, becomes significant during adulthood at the age of 25 and thereafter remains essentially constant throughout the ages. (5) Finally, the sporadic counterparts to the CRCs diagnosed in the setting of Lynch syndrome very curiously develop almost exclusively in the right colon, whereas this right-sidedness is much less pronounced in Lynch syndrome. To explain these anomalies, we propose a model of RER+ carcinogenesis based on the simple idea that the RER mutator phenotype has only short-term viability in normal cells. The proposed model states that the RER+ carcinogenesis is divided into two clearly distinct evolutive phases: (1) a preliminary phase starting with the counter-selective loss of mismatch repair function, in which most clones with the RER mutator phenotype are eliminated through apoptosis or an accelerated aging process; (2) an explosive phase that is initiated only if mutations blocking apoptosis and senescence, rapidly acquired during the short life span of the nontransformed RER+ clones, eventually rescue one RER+ cell that gives rise to the malignant clone. It will be shown that this theoretical framework with its heterodox initiation process not only possesses the virtue of allowing an understanding of Lynch syndrome, but may also have broader applications to all research fields dealing with carcinogenesis.