In 2006 and 2007, the SPHERES testing facility, a series of micro-satellites developed by the MIT Space Systems Laboratory and designed to operate autonomously inside the International Space Station, has been extensively used to demonstrate multiple autonomous docking scenarios. Following the first autonomous docking to a tumbling target ever achieved on-orbit and presented at a conference last year, some off-nominal docking scenarios were attempted. Multiple autonomous docking maneuvers were achieved in the presence of simulated failures. In some situations, the chaser was required to perform a collision avoidance maneuver, which consisted in a simple thrust sequence that brought it on a safe trajectory. A docking maneuver was also attempted with a satellite tumbling such that its docking port swept a plane that does not include the chaser’s initial position, to simulate a servicing satellite docking to an uncontrollable target. One last scenario was an attempt to dock to a target satellite with its docking port sweeping a cone. Two different algorithms were used to perform some of these experiments. The first one involved a traditional glideslope approach, where the chaser is commanded to reduce its approach velocity with the distance-to-go. The second involved a pre-planned trajectory ensuring a passive abort in case a failure is detected. Following the success of these experiments, interesting conclusions were made on the performance of both algorithms, based on the observations during the experiments and the telemetry. This paper will present the results of these experiments, along with a comparison of the performance of both algorithms for the different scenarios attempted.