AbstractA Bayesian analysis is made of the probability of core melt (defined as“core on the floor”) for U.S. reactors, using methods that are an extension of those used in an earlier work, and that can therefore be used to answer different questions. The essential new point is that at any time the estimate of the probability of core melt varies continuously as melt-free experience accumulates, with a substantial effect on the estimates. The most interesting result is that the probability of no-melt-yet stays large somewhat longer than one might have guessed, largely because melt-free experience feeds on itself in reducing the probability of subsequent core melt, and this process continues until the first melt.