Abstract. The immune system uses at least 108 randomly generated “detectors” to identify potentially harmful pathogens. We here consider the detectors that are formed by the “random” T cell receptors (TCR) on circulating naive T cells, with each cell typically expressing a single type of TCR. Human naive T cells are long-lived (several years), divide infrequently, compete for space and survival factors, and are rarely triggered to respond by an invading pathogen. Novel naive T cell clonotypes, i.e. cells expressing a new random TCR, can only be produced in the thymus, which is an organ that shrinks and ultimately disappears when we age. This begs the question om how such a diverse “ecosystem” of competing naive T cell populations is maintained over time, and how rapidly TCR repertoire diversity erodes during aging. We analyze deep-sequencing data of TCRs in volunteers to characterize the frequency distribution of different TCRs in the repertoire, and develop a neutral competition model similar to the ecological model of Stephen Hubbell to describe the data. Surprisingly, we find that a strong relation between the probability with which a TCR is expected to be generated in the thymus, and the frequency of that TCR in the naive T cell repertoire. It is not understood why a small fraction of TCRs should be so abundant in the repertoire, but this does suggest that the competition is neutral, and helps to explain the slow erosion of repertoire diversity when we age.