In the Drosophila melanogaster mutant sevenless, the R7 photoreceptor in each of the repeating units, or ommatidia, of the compound eye fails to form. We have determined the nucleotide sequence and structure of the sevenless transcription unit. Overlapping cDNA clones from an eye imaginal disc library were isolated, and these together with corresponding genomic regions were sequenced. The positions of the major and two minor transcription start sites were mapped. The gene encodes a putative cell-surface receptor, the unmodified form of which is predicted as 288 kD, bearing a cytoplasmic tyrosine kinase domain. Several structural features distinguish sevenless from other tyrosine kinase receptors, most notably the large size of its extracellular domain. Moreover, unlike the insulin or epidermal growth factor (EGF) receptors, a putative amino-terminal signal sequence does not immediately follow the initiating methionine codon. Rather, a 21-amino-acid hydrophobic sequence is found 56 amino acids after the likely initiating methionine codon, suggesting that this sequence may function as an amino-terminal anchor. If so, the sevenless protein would be expected to have its 2001-amino-acid extracellular domain anchored as a loop at either end in the membrane. Finally, we have generated a number of transformant lines that carry either a 16.3-kb genomic fragment that extends approximately 950 bp upstream of the transcription start sites or constructs in which a subset of the introns present in the genomic sevenless DNA have been removed. The degree of rescue of the sevenless mutant phenotype has been assayed in these lines using morphological and behavioral assays.