A series of wild-type and mutant raf genes was transfected into NIH 3T3 cells and analyzed for transforming activity. Full-length wild-type c-raf did not show transforming activity. Two types of mutations resulted in oncogenic activity similar to that of v-raf: truncation of the amino-terminal half of the protein and fusion of the full-length molecule to gag sequences. A lower level of activation was observed for a mutant with a tetrapeptide insertion mapping to conserved region 2 (CR2), a serine- and threonine-rich domain located 100 residues amino-terminal of the kinase domain. To determine essential structural features of the transforming region of raf, we analyzed point and deletion mutants of v-raf. Substitutions of Lys-56 modulated the transforming activity, whereas mutation of Lys-53, a putative ATP binding residue, abolished it. Deletion analysis established that the minimal transforming sequence coincided precisely with CR3, the conserved Raf kinase domain. Thus, oncogenic activation of the Raf kinase can be achieved by removal of CR1 and CR2 or by steric distortion and requires retention of an active kinase domain. These findings are consistent with a protein structure model for the nonstimulated enzyme in which the active site is buried within the protein.