Two experiments examined the nature of recognition memory by asking how subjective reports of remembering change over time. In Experiment 1, participants were asked to report their experience of remembering using the well-known remember–know–guess procedure. Estimates of recollection declined over a 14-day period, but estimates of familiarity remained constant, suggesting that the processes are independent. In Experiment 2, participants were asked to report their confidence in their recognition decisions. Subjective reports of confidence were analysed via receiver operating characteristics and also indicated different rates of decline for recollection and familiarity. Superficially, the data appear to support a dual-process account of recognition, but close inspection shows the data to be consistent with a simple signal detection model. The conclusion is that although the phenomenal experience of remembering changes over time this is most likely to be predicated on a single process.