Abstract The evolution of simple satellite DNAs was examined by DNA-DNA hybridization of ten Drosophila melanogaster satellite sequences to DNAs of the sibling species, Drosophila simulans and Drosophila erecta. Seven of these repeat types are present in tandem arrays in D. simulans and each of the ten sequences is repeated in D. erecta. In thermal melts, six of the seven satellite sequences in D. simulans and seven of the ten sequences in D. erecta melted within 1 deg.C of the corresponding values in D. melanogaster. The remaining sequences melted within 3 deg.C of the homologous hybrids. Therefore, there is little or no alteration in those satellite sequences held in common, despite a period of about ten million years since the divergence of D. melanogaster and D. simulans from a common ancestor. Simple satellite sequences appear to be more highly conserved than coding regions of the genome, on a per nucleotide basis. Since multiple copies of three satellite sequences could not be detected in D. simulans yet are present in D. erecta, a species more distantly related to D. melanogaster than is D. simulans, these sequences show discontinuities in evolution. There were major quantitative variations between species, showing that satellite DNAs are prone to massive amplification or diminution events over timespans as short as those separating sibling species. In D. melanogaster, these sequences amount to 21% of the genome but only 5% in D. simulans and 0.4% in D. erecta. There was a general trend of lower abundance with evolutionary distance for most satellites, suggesting that the amounts of different satellite sequences do not vary independently during evolution.