To determine the fate of an episomally expressed transgene, mouse, rat, and cow zygotes were injected into the perivitelline space with approximately 100 pl of buffer containing the replication-defective human adenovirus, AdCMVLacZ/sub360. Viral concentrations ranged from 2.5 to 2.5 x 105 plaque-forming units (pfu)/100 pl. As viral titer increased, fewer embryos were able to develop to blastocysts. In the mouse, the percentage of blastocysts formed ranged from 82% in controls to 16% after injection at the highest titer. In the rat and cow, a similar decrease in blastocyst formation was noted (62% to 6% and 26% to 4%, respectively). Reporter gene (galactosidase, LacZ) activity could be detected in mouse embryos after injection at a concentration of only 25 pfu/100 pl, whereas a tenfold higher titer was required in the other two species to observe the blue LacZ reaction product. When examined after 5 (mouse), 6 (rat), or 9 (cow) days of in vitro culture, the proportion of LacZ-positive embryos ranged from 15% to 96%, 6% to 76%, and 18% to 58% in mouse, rat, and cow embryos, respectively, depending upon viral concentration. However, a large percentage of positive embryos proved to be expression mosaics, the degree of which was likewise dependent on titer. While none of the embryos showed LacZ activity at 30 h after injection, 70% of mouse, 8% of rat, and 20% of cow embryos expressed the reporter gene at 42 h. Delaying the timing of injection revealed that the efficiency with which mouse and rat embryos could be infected decreased with increasing degree of differentiation. Only 35% and 18% of mouse embryos expressed the reporter gene after injection at the morula or blastocyst stage, respectively. A similar drop in efficiency was noted in rat embryos when injections took place at the 8-cell, morula, or blastocyst stage, with 70%, 33%, and 9% of embryos, respectively, subsequently showing LacZ activity. Likewise, advanced development resulted in a decrease in the efficiency of viral-mediated gene transfer in cow embryos, with 100%, 78%, and 68% of embryos being positive after injection at the 8-cell, morula, or blastocyst stage, respectively. These results demonstrate that a human adenovirus can be used to express a reporter gene transiently in nonhuman embryos.