Early attempts to generate new restriction specificities by recombination between allelic restriction-modification systems have been unsuccessful. Bullas et al. succeeded in isolating a new specificity, SQ, in Salmonella that they interpreted as being the result of a recombination event between the parental strains, Salmonella typhimurium and S. postdam, which encode the SB and SP restriction systems, respectively. This interpretation has recently been confirmed by DNA heteroduplex studies with the SB, SP and SQ structural genes. We have determined the DNA sequences recognized by the SB and SP enzymes and found that, like all type I restriction sequences, they are split into two specific domains by a spacer of nonspecific sequence that, for both SB and SP, is 6 base pairs (bp) long. We have now determined the sequence recognized by the recombinant SQ enzyme and find that it is a hybrid between the SB and SP sequences, containing one specific domain from each parental strain. This result implies that each of the two specific domains is recognized by a physically distinct part of the enzyme.