Retinae from embryonic rats transplanted over the midbrain of newborn host rats establish connections with visual centres of the host brain, which mediate a pupilloconstrictor response in the host eye when the transplant is stimulated by light. The changes in the size of the host pupil can be measured accurately with a pupillometry system. We have taken advantage of the additional observation that while grafts between rat strains, as between Long Evans and Sprague-Dawley strains, may survive indefinitely, they can be induced to reject by skin grafting from the strain providing the donor retinal tissue. Combining pupillometry with skin grafting provides a useful way of examining correlated anatomical and behavioural changes associated with graft rejection from its earliest stage to the point of overt destruction. Even within three days of skin grafting, the amplitude and speed of constriction as well as the response latency all showed significant enhancement from normal, and this was sustained for a further week or more. Response deterioration followed during the second week post-skin grafting, but the exact timing varied considerably among animals. Anatomical observations of the process of retinal rejection showed the first invasion of lymphocytes to occur between days 5 and 7 and total degeneration of the retinal transplant and its projections to occur by two to three weeks post-skin grafting. The lymphocytic infiltration was preceded by upregulation of microglia, which expressed both class I and II major histocompatibility antigens and by activation of astrocytes identified by their expression of glial fibrillary acidic protein. Within the target region of retinal transplant axons, major histocompatibility antigen expression and astrocytic responses preceded degeneration of transplant derived axons (demonstrated by the Fink-Heimer stain) and there was no evidence for any lymphocytic infiltration during transplant rejection. These observations show that the earliest stages of microglial activation are accompanied by an enhancement of response parameters, but that the functional failure finally occurs only at an advanced stage of graft destruction. The absence of lymphocytic infiltration into areas receiving terminals from axons of transplant origin, even though these contain significant numbers of reactive microglia, suggests that the terminal axonal processes are not a primary target for the immune response.