Retroviruses package two copies of genomic RNA into viral particles. During the minus-sense DNA synthesis step of reverse transcription, the nascent DNA can transfer multiple times between the two copies of the genome, resulting in recombination. The mechanism for this process is similar to the process of obligate strand transfers mediated by the repeat and primer binding site sequences. The location at which the DNA 3 terminus completely transfers to the second RNA strand defines the point of crossover. Previous work in vitro demonstrated that reverse transcriptase pausing has a significant impact on the location of the crossover, with a proportion of complete transfer events occurring very close to pause sites. The role of pausing in vivo, however, is not clearly understood. By employing a murine leukemia virusbased single-cycle infection assay, strong pausing was shown to increase the probability of recombination, as reflected in the reconstitution of green fluorescent protein expression. The infection assay results were directly correlated with the presence of strong pause sites in reverse transcriptase primer extension assays in vitro. Conversely, when pausing was diminished in vitro, without changing the sequence of the RNA template involved in recombination, there was a significant reduction in recombination in vivo. Together, these data demonstrate that reverse transcriptase pausing, as observed in vitro, directly correlates with recombination during minus-sense DNA synthesis in vivo.Retroviruses exhibit a high frequency of homologous recombination (7,29,35,36,37,43,58,59,61,65). Human immunodeficiency virus type 1 (HIV-1) recombination occurs at a rate of approximately 2.4 ϫ 10 Ϫ4 /bp per replication cycle or higher while the direct repeat deletion rate for murine leukemia virus (MLV) was reported at 1.09 ϫ 10 Ϫ5 /bp per replication cycle (29,46,50). The rate of nonhomologous recombination is much lower but is still mediated by short homologous sequences (62, 64). During one round of reverse transcription, multiple crossover events often appear to occur at a significantly higher rate than mathematically anticipated, suggesting that a second crossover is not an independent event (negative interference) (2, 3). There are conflicting reports concerning this phenomenon in HIV-1; early work suggested the existence of high negative interference (29), while a more recent report obtained the expected distribution of multiple crossover events (51).Recombination of two different retrovirus strains relies on coinfection of a host cell by two genetically distinct viruses and copackaging of one genomic RNA from each integrated provirus into a single virion (26,27,55). The heterozygous virion then infects a new target cell where a chimeric recombinant provirus can be generated during reverse transcription. Recently, copackaging of heterozygous virions was demonstrated to be strongly affected by the identity of the dimerization initiation sequence. Genomes with different dimerization initiation sequences were packaged...