Phase space density profiles for protons and electrons with first invariants &mgr;≤100 MeV/G (integral invariants K=0.3 and 0.6 G1/2 RU) previously derived from measurements by the low energy charged particle (LECP) detector on Voyager 2 during the 1986 Uranus encounter are analyzed using solutions of the time-averaged radial diffusion equation in a dipolar magnetic field. These profiles are selected for their consistency with an absence of local sources of particles. A loss model consisting of absorption by the major inner satellites Miranda, Ariel, and Umbriel is assumed and the corresponding form of the time-averaged radial diffusion coefficient D(L) (taken to be of the form D(L)=D0Ln, where n is an integer) is determined by a minimum-variance fit to the phase space density profiles. Satellite macrosignatures present in the experimentally derived profiles are approximately reproduced in several cases lending credence to the loss model and indicating that magnetospheric distributed losses are not as rapid as satellite absorption near the minimum satellite L shells for these particles. The latter inference implies an upper limit of approximately 10 cm-3 for the neutral hydrogen number density near the orbit of Ariel, based on a comparison of charge exchange lifetimes with the calculated satellite absorption lifetimes. The preferred forms for D(L) are characterized by a low-order L dependence (~L3-L4) and an amplitude (D0≂10-11--10-10 R2U -1). The inferred L dependence is least consistent with ''terrestrial-type diffusion mechanisms including magnetic impulses and electrostatic field fluctuations of purely magnetospheric origin. ''Jovian-type mechanisms including the ionospheric dynamo electric field mechanism and the centrifugal interchange instability in regions of strong negative radial plasma density gradients predict L dependences that are more consistent with the inferred form. Of these, the ionospheric dynamo mechanism is most viable at Uranus. ¿ American Geophysical Union 1989