In this paper we evaluate quantitatively the hypothesis that topographic modification of floor-fractured craters on the moon was accomplished by viscous relaxation. Adopting the simple assumption that the moon may be modeled as having a uniform Newtonian viscosity, we compare the observed topographic profiles for a number of floor-fractured craters with the profiles predicted from the viscous relaxation of topography of fresh craters of similar diameter. Despite the simplicity of the rheological model, the comparison is quite good. The floor uplift, the rim subsidence, and the apparent subsidence outside the rim for the several floor-fractured craters considered are well matched by the viscous relaxation hypothesis. Floor fractures, while indicating that a purely viscous model is not strictly valid, can be explained by the effects of isostatic adjustment on a thin brittle lithosphere. The association of many floor-fractured craters with impact basins and with the time of mare volcanism can be understood in terms of a pronounced acceleration of crater relaxation in local regions of anomalously high near-surface temperatures and therefore of low effective viscosity and thin lithosphere. The quantitative extent of relaxation of floor-fractured craters can be interpreted in terms of a limited time interval of substantial relaxation for each crater. That time interval ended for each crater after local cooling had been sufficient for the viscosity to rise, for the lithosphere to thicken, and for the present topographic relief to be 'frozen in.' Thus viscous relaxation is a viable hypothesis to explain the topographic profiles of a number of lunar floor-fractured craters, and the extent to viscous relaxation of crater topography may serve as a tool to map lateral and temporal variations in the shallow thermal structure of the moon and other planets and satellites.