The targeting of DNA integration in retrovirus-infected cells is a central yet very poorly understood aspect of the biology of the virus. To investigate this problem, we have assessed the use of specific sites for integration targets of avian leukosis virus (ALV) DNA within defined regions of turkey embryo fibroblast (TEF) cellular DNA. For this purpose, we developed an assay of sufficient sensitivity and specificity to allow detection and location of single integration events in a population of 5 million cells. Targets selected for study were either regions cloned by virtue of a previous integration event or clones chosen at random from cellular DNA. By use of this approach, we found that all genomic regions tested contained integration targets, with a frequency that varied from approximately 0.2 to 4 times that expected for random integration. Within regions, the frequency of use of specific sites varied considerably, with some sites used up to 280 times random frequency. When one region was introduced into cells at moderately high copy number by transfection, it provided integration targets in a pattern very much like that seen with the same sequence in vitro. On the basis of our sampling, we conclude that most or all regions of the TEF genome are accessible to ALV retroviral integration. As with integration in vitro, integration specificity seems to be determined largely by local structural features rather than accessibility of specific regions.