In humans, the amount of terminal (TTAGGG)n, telomeric DNA decreases during aging of various somatic cell types in vitro and in vivo. While the factors accounting for telomere shortening have not been thoroughly established, the inability of the DNA replication machinery to completely copy chromosomal termini (the "end replication problem") and the absence in somatic cells of telomerase, the enzyme that synthesizes telomeric DNA de novo, is a likely mechanism. One prediction of this hypothesis is that telomere shortening should be dependent on cell division. Thus we analyzed telomere length in actively dividing and quiescent cells in vitro and in vivo. In circular outgrowths of cultured human diploid fibroblasts (HDF), cells at the outer periphery had a significantly lower mean terminal restriction fragment (TRF) length (P = 0.011) and telomeric signal intensity (P = 0.024) than cells at the center. Also, the rate of telomere shortening over time for HDFs held quiescent was not statistically significant (m = -12 bp/day, P = 0.16) while that for serially passaged cells was significant (m = -34 bp/day, P = 0.017). To examine the rate of telomere shortening for quiescent cells in vivo, we measured mean TRF length in brain tissue from adult donors ranging in age from 32-75 years. No significant decrease was observed as a function of donor age (P = 0.087), in contrast to the shortening of telomere length that occurs during in vivo aging of mitotically active cells (P = 0.0001). These observations show that telomere shortening is largely, if not entirely, dependent on cell division and support the end replication problem as a mechanism for this process and the use of telomere length as a biomarker for replicative capacity.