The attrition of telomeres, the ends of eukaryote chromosomes, is thought to play an important role in cell deterioration with advancing age. The observed variation in telomere length among individuals of the same age is therefore thought to be related to variation in potential longevity. Studies of this relationship are hampered by the time scale over which individuals need to be followed, particularly in long-lived species where lifespan variation is greatest. So far, data are based either on simple comparisons of telomere length among different age classes or on individuals whose telomere length is measured at most twice and whose subsequent survival is monitored for only a short proportion of the typical lifespan. Both approaches are subject to bias. Key studies, in which telomere length is tracked from early in life, and actual lifespan recorded, have been lacking. We measured telomere length in zebra finches (n = 99) from the nestling stage and at various points thereafter, and recorded their natural lifespan (which varied from less than 1 to almost 9 y). We found telomere length at 25 d to be a very strong predictor of realized lifespan (P < 0.001); those individuals living longest had relatively long telomeres at all points at which they were measured. Reproduction increased adult telomere loss, but this effect appeared transient and did not influence survival. Our results provide the strongest evidence available of the relationship between telomere length and lifespan and emphasize the importance of understanding factors that determine early life telomere length.