Abstract The collapse of oyster populations and the fisheries they support has been a worldwide phenomenon, but studies of oyster demography in situ prior to and after the collapse have been rare. We used time series of stage-based counts of eastern oysters Crassostrea virginica in Apalachicola Bay, Florida, to help understand how abundance and demographic rates may have changed in the decade after the 2012 collapse relative to the period before the collapse. We relied on a Bayesian hierarchical model in which the latent stage structure of the oyster population (i.e., densities of spat, sublegal oysters, and legal oysters) was governed by a system process and where the count data represented summaries of that latent structure. Count data were sufficient to conduct this on two large oyster bars that had some of the highest precollapse oyster densities. We also examined nine other bars with less data for any temporal trends in postcollapse abundance that might be associated with recent restoration efforts. Among the 11 bars examined, oyster densities were often increasing prior to the collapse and were very low, without detectable trends, afterward. Based on our demographic analyses, mortality rates of Apalachicola Bay oysters in the decade after the collapse generally exceeded (often greatly so) those during the precollapse period for all oyster stages. On the other hand, spat settlement rates apparently were increasing prior to the collapse and remained high during the postcollapse period. Simulations of postcollapse demography suggest that without improved survival rates, further declines of the oyster population can be expected. We discuss these findings in light of ongoing restoration and management efforts and suggest ways in which rapid transitions to undesirable socio-ecological regimes might be avoided in the future.