Two distinct site-specific retrotransposon families, named RT1 and RT2, from the sibling mosquito species Anopheles gambiae and A. arabiensis, respectively, were previously identified. Both were shown to occupy identical nucleotide positions in the 28S rRNA gene and to be flanked by identical 17-bp target site duplications. Full-length representatives of each have been isolated from a single species, A. gambiae, and the nucleotide sequences have been analyzed. Beyond insertion specificity, RT1 and RT2 share several structural and sequence features which show them to be members of the LINE-like, or non-long-terminal-repeat retrotransposon, class of reverse transcriptase-encoding mobile elements. These features include two long overlapping open reading frames (ORFs), poly(A) tails, the absence of long terminal repeats, and heterogeneous 5' truncation of most copies. The first ORF of both elements, particularly ORF1 of RT1, is glutamine rich and contains long tracts of polyglutamine reminiscent of the opa repeat. Near the carboxy ends, three cysteinehistidine motifs occur in ORF1 and one occurs in ORF2. In addition, each ORF2 contains a region of sequence similarity to reverse transcriptases and integrases. Alignments of the protein sequences from RT1 and RT2 reveal 36% identity over the length of ORF1 and 60%o identity over the length of ORF2, but the elements cannot be aligned in the 5' and 3' noncoding regions. Unlike that of RT2, the 5' noncoding region of RT1 contains 3.5 copies of a 500-bp subrepeat, followed by a poly(T) tract and two imperfect 55-bp subrepeats, the second spanning the beginning of ORF1. The pattern of distribution of these elements among five sibling species in the A. gambiae complex is nonuniform. RT1 (43), and the absence of LTRs indicates a very different replication strategy that is not well understood. Peptide alignments show that the RT and capsid-like domains of non-LTR retrotransposons are more closely related to one another than to those of other RTencoding mobile elements (23,51,66,68). However, although the coding region may occupy over 80% of the length of a given element, the sequence diversity among characterized non-LTR retrotransposons is high enough to preclude amino acid alignment of any but very short segments.Many non-LTR retrotransposons, such as mammalian Li elements (32) and Ti elements from mosquitoes in the Anopheles gambiae complex (8, 9), are widely dispersed in the host genomic DNA and have no apparent insertion site specificity. However, some elements of this same class exhibit a preference for a particular target site. In three different trypanosomatid protozoa, three distinct elements (CRE1, SLACS, and CZAR) interrupt some portion of the array of spliced leader RNA genes at the same conserved sequence (2). The Drosophila G element occupies a precise site in the intergenic spacers of the rRNA genes (22