All wild-type isolates of Epstein-Barr virus (EBV) analyzed to date for allelic polymorphisms of the nuclear antigen EBNA2 gene (in the BamHI YH region of the genome) and of the EBNA3A,-3B, -3C genes (tandemly arranged in the BamHI E region) have proved either uniformly type 1 or uniformly type 2 at all four loci. The absence of detectable intertypic recombination in the wild probably reflects the rarity with which individual carriers, and certainly individual target cells, become coinfected with both virus types. Studying a group of human immunodeficiency virus-positive T-cell-immunocompromised patients known to be at enhanced risk of multiple EBV infections, we have isolated intertypic EBV recombinants from 2 of 40 patients analyzed. These recombinants, whose in vitro transforming capacity appeared at least equal to that of type 1 strains, carried a type 1 EBNA2 allele and type 2 EBNA3A,-3B, and -3C alleles. This was clearly demonstrable at the DNA level by PCR amplification using type-specific primer-probe combinations and was confirmed at the protein level (for EBNA2 and EBNA3C) by immunoblotting with type-specific antibodies. In one patient, the recombinant appeared to be the predominant strain, being the virus most commonly rescued by in vitro transformation both from the blood and from the throat washings on two separate occasions 20 months apart. A regular type 1 virus strain was also present in this individual, but this was not related to the recombinant since the two viruses carried type 1 EBNA2 genes with different patterns of variance from the B95.8 prototype sequence. In the other patient, recombinants were isolated on one occasion from the blood and on a separate occasion, 21 months later, from the throat; these recombinants were almost certainly related, being identical at several genomic polymorphisms and differing only in one facet of the "EBNAprint," the size of the EBNA1 protein. Three different type 1 viruses were also isolated from this patient, two of which carried EBNA2 genes with the same pattern of sequence variation from B95.8 as the recombinant; however, since this is a fairly common pattern of variance, the relationship of these viruses to the recombinant remains an open question. We infer that intertypic recombinants of EBV are not uncommon in HIV-positive T-cell-immunocompromised patients, that they arise in such individuals as a consequence of their increased frequency of mixed-type infections, and that they will prove capable of efficient transmission in the human population.