Little is known about the mechanism of poxvirus recombination even though construction of recombinant viruses by recombination-dependent methods is a widely adopted technique. We have shown previously that transfected DNAs are efficiently recombined while replicating in cells infected with Shope fibroma virus. Because recombinant DNA can be recovered from infected cells as a high-molecular-weight head-to-tail concatemer, it was possible to transfect genetically marked lambda DNAs into infected cells and assay recombinants as bacteriophage particles following in vitro packaging. This approach was used in this study to examine how marker distance and marker orientation influence recombination in Shope fibroma virus-infected cells. Simple two-factor crosses were readily modelled by using a mapping function derived from classical phage studies and showed low negative interference (I = -2.8 +/- 0.5) in crosses involving markers greater than 100 bp apart. More complex four- and five-factor crosses showed that the recombination frequency per unit distance was not constant (rising as the marker separation was reduced from 100 to 1 bp) and that crosses performed in poxvirus-infected cells are subject to high negative interference. One consequence is that marker orientation does not dramatically influence the outcome of most Shope fibroma virus-catalyzed crosses in clear contrast to what is observed in adenovirus or simian virus 40-infected cells. These results can be interpreted to indicate that similar statistical and physical constraints influence both viral and phage recombination and suggest that heteroduplexes may be important intermediates in the poxvirus recombination process.