We have shown previously that the putative oncogene pim-1 is frequently activated by provirus insertion in murine leukemia virus-induced T cell lymphomas. Here we describe the structure of the pim-1 gene as determined by sequencing genomic and cDNA clones. The gene has an open reading frame, encoding a protein of 313 amino acids, extending over six exons and preceded and followed by stop codons in all reading frames. Proviruses always integrate outside the protein-encoding domain, showing a high preference for a small region in the 3'-terminal exon; integration in the 3' exon results in relatively high levels of pim-1 mRNA. Computer search reveals homology between pim-1 and protein kinases: all the domains characteristic of protein kinases are conserved in the pim-1 amino acid sequence. The highest homologies were observed with the protein-serine kinases.