The major goal of these experiments is to derive stable, helper independent, retroviral vectors using the SR-A strain of Rous sarcoma virus. Because src is flanked by direct repeats of 110 bases, both src, and sequences that replace src in vector constructions, are lost at high frequency. We have sought to eliminate this homology in order to stabilize the vectors. One copy of the direct repeat must be retained for the virus to replicate properly. Because the downstream direct repeat is linked to the polypurine tract the entire downstream direct repeat cannot easily be eliminated. We therefore sought to eliminate the upstream direct repeat. Using linkers a series of defined deletions and duplications has been created within the region between env and src. The region is relatively large, 379 bases, and has a complex history (it is derived from three different nucleic acid segments each with a distinct and separate origin). We show here that this region provides no functions essential for growth and, for src expression, provides only a functional splice acceptor. We were able to successfully replace the splice acceptor found in the wild type virus with an unrelated splice acceptor partially derived from a synthetic DNA segment. The final product is a replication competent virus that expresses src, and that lacks the entire upstream repeat. Since src is flanked by ClaI sites in these constructions, src can easily be replaced by other genes. Substituting the Tn5 neo gene for src in this construction yields a virus that expresses the neo gene nonselectively.