Telomeres shorten with age, which may be linked to genomic instability and an increased risk of cancer. To explore this association, we analyzed telomere length in normal colorectal tissue of individuals at different ages using quantitative-fluorescence in situ hybridization (Q-FISH) and quantitative-PCR (Q-PCR). Using Q-FISH, we also examined the histologically normal epithelium adjacent to, or distant from, colon adenomas and cancers, in addition to the neoplasms. Q-FISH and Q-PCR showed that telomere length was inversely associated with age until approximately ages 60 to 70; surprisingly, beyond this age, telomere length was positively associated with age. This association was found exclusively in epithelial, and not in stromal, cells. Peripheral blood lymphocytes showed an inverse association between telomere length and age, but without any apparent increase in telomere length in the oldest individuals. Telomere length in larger adenoma lesions (>2 cm) was significantly shorter than in normal adjacent (P = 0.004) or normal distant (P = 0.05) tissue from the same individuals. However, telomere length in histologically normal epithelium adjacent to cancers or in adenomas <2 cm was not statistically different from that of the normal distant mucosa or from normal controls, evidence that a telomere-shortening field effect was not present. We suggest that the positive association between telomere length and age in the oldest patients is a consequence of selective survival of elderly patients with long colonocyte telomeres.