The dipteran Chironomus tentans has complex tandemly repeated 350-bp DNA sequences at or near the chromosome ends. As in Drosophila melanogaster, short simple repeats with cytosines and guanines in different strands have never been observed. We were therefore interested in learning whether the Chironomus repeats could have evolved from simple sequence telomeric DNA, which might suggest that they constitute a functional equivalent. We screened for repeat units with evolutionarily ancient features within the tandem arrays and recovered two clones with a less-evolved structure. Sequence analysis reveals that the present-day 350-bp unit probably evolved from a simpler 165-bp unit through the acquisition of transposed sequences. The 165-bp unit contains DNA with a highly biased distribution of cytosine and guanine between the two strands, although with the ratios inverted in two minor parts of the repeat. It is largely built up of short degenerate subrepeats for which most of the sequence can be reconstructed. The consensus for the subrepeat sequence is similar to the simple telomeric repeat sequences of several kinds of eukaryotes. We propose that the present-day unit has evolved from telomeric, simple sequence, asymmetric DNA from which it has retained some original sequence features and possibly functions.At the ends of most eukaryotic chromosomes, there are short, simple DNA repeats, slightly different in various lower and higher eukaryotes (reviewed in reference 5). Such DNA of different origin has one feature in common: all or most of the cytosines are in one strand and the guanines in the other. The terminal DNA can be added to the ends enzymatically by telomerase (17,18) (8,9,24,29). Similar DNA has also been described for some Drosophila species (1). Because of the well-defined structure of the 350-bp TA repeats in the subgenus Camptochironomus, with two pairs of subrepeats in an alternating arrangement, separated by four linker regions, nonrepetitive within the 350-bp unit, the TA repeats provide an interesting model to learn how a complex unit is formed during evolution from a simpler organization. We have earlier asked whether the formation of subrepeats, which show more than 90% mutual homology, is due to differences in rates of evolution along the 350-bp repeat unit after previous duplication of an approximately 175-bp half-repeat. Alternatively, forces exist * Corresponding author.that give members of a pair of subrepeats a parallel evolution. We have compared repeat units in the closely related Chironomus tentans and C. pallidivittatus and concluded that mutations are preferentially located in linker regions, thus supporting the first alternative (24).From the interspecies comparison, we learned what happened after speciation, when most of the morphology of the repeat unit is already established. For present purposes, we searched for a repeat unit with early evolutionary characters. For tandem repeats, this is not necessarily unrealistic, since tandem repeats may contain units at their ends representi...