Diode laser photocoagulation was applied to rabbit retina simulating scatter treatment using an endolaser probe and in a manner simulating treatment of peripheral retinal breaks using a transscleral retinopexy probe. Clinically appearing mild, moderate, and severe burns were created by altering the burn duration in one eye and by altering the power setting in the fellow eye. Histopathologic results demonstrated the clinically evident dose-response effect with sparing of inner retinal cellular elements with mild burns and full-thickness retinal cell loss with severe burns. Bruch's membrane ruptures were seen in three of 42 endophotocoagulation severe spots placed with high power, but in none of the 42 severe spots placed with long burn duration. Thus, longer burn duration appeared to be a safer way to produce a severe burn than higher power. Burns characteristically bloomed during the several seconds following laser application by both modalities, possibly indicating a deep source of energy absorption. Scleral effects, seen only when high energy levels were used to treat atrophic areas, were mild.