The theory of satellite loss resulting from a giant impact on Uranus (Parisi and Brunini 1997, Planer. Space Sci. 45, 181-187) is revisited, in the light of the discovery of its five outer moons (Gladman et al. 1998, Nature 392, 897-899; Gladman et al. 2000, Icarus 147, 320-324; erratum 148, 320). Physical conditions and dynamical constraints in the great collision scenario and restrictions in the possible mechanisms for the origin of the outer uranian satellites are obtained from the knowledge of their actual orbital properties. We conclude that the existence of these moons implies that their origin must be connected to a breakup process. Other scenarios for their origin cast serious doubts on the occurrence of a giant collision at the end of Uranus' formation process to account for its large spin axis obliquity. (C) 2002 Elsevier Science (USA).
