To prevent their recognition as DNA breaks, the ends of linear chromosomes are organized into telomeres, which are made of proteins bound to telomere-specific, double-stranded repeats and to single-stranded DNA extensions, the G-tails. The mammalian heterogeneous nuclear ribonucleoparticule A1 and A2 proteins can bind with high affinity to such G-tails. Moreover, previous work established that in certain mouse cells a severe reduction in the level of A1 is associated with shortened telomeric repeat tracts, and restoring A1 expression increases telomere length. Here, we document that the expression of A1/A2 proteins is elevated in a variety of human cancers, whereas A1/A2 expression is lower or absent in normal tissues. To determine whether the status of A1/A2 proteins could be improved from cancer markers to cancer targets, we used small interfering RNA-mediated RNA interference to elicit a reduction in A1/A2 proteins in a variety of human cell lines. We show that this treatment provoked specific and rapid cell death by apoptosis in cell lines derived from cervical, colon, breast, ovarian, and brain cancers. Cancer cell lines that lack p53 or express a defective p53 protein were equally sensitive to a small interfering RNA-mediated decrease in A1/A2 expression. The reduction in A1/A2 levels in HeLa cells was associated with a change in the distribution of the lengths of G-tails, an event not observed when apoptosis was induced with staurosporine. Remarkably, comparable decreases in the expression of A1/A2 in several mortal human fibroblastic and epithelial cell lines did not promote cell death. Thus, manipulating the level and activity of A1/A2 proteins may constitute a potent and specific approach in the treatment of human cancers of various origins.