We investigated the traffic of allograft-responding leukocytes between the host and graft without handling of these cells in vitro. The blood flow between the host and graft was disconnected, the proliferating cells were labeled with [3H]thymidine selectively in the graft or in the host, the label was chased with cold thymidine, and the circulation was reestablished. The localization of labeled cells was quantitated by autoradiography. The first host-derived labeled cells appeared in the graft and graft-derived labeled cells in the host, already on the 1st d after transplantation. This was followed by an exponential increase in the labeled cell traffic in both directions. The peak of traffic was observed on day 4 after transplantation, whereafter the traffic rapidly declined and tapered off. This decline was not due to exhaustion of supply, as the labeled cells continued to proliferate in their original compartments, nor to a slowdown of blood circulation, which took place 2-3 d later. We consider the decline to indicate that the rejection has proceeded to a (irreversible) stage autonomous of the host lymphatic and hematopoietic system. During the exponential increase, nearly one-third of the graft-infiltrating inflammatory cells were replaced as a consequence of relocalization during each 18-h-period. All mononuclear white cell types, with the exception of granulocytes, participated in the traffic. Most lymphoid cells entrapped in the graft were descendents of recent cell divisions; most of the mononuclear phagocytes derived from a preexisting phagocyte pool. The entrapment of labeled leukocytes in a relevant graft was specific: when an allograft and an autograft were simultaneously transplanted, a more than 50-fold entrapment was observed in the allograft, compared with the autograft. Very few of the cells localized in irrelevant positions, such as the liver and lung, of the recipient.