Although the Morris water task has been used in hundreds of studies of place learning, there have been no systematic studies of retention of the place memory. We report that retention, as measured by selective search behavior on a probe trial, is excellent when the retention interval is short (5-10 min). However, performance rapidly deteriorates, so that by approximately 4 h the search is no longer selective. Additional experiments revealed that selective search at longer retention intervals was improved by inserting gaps between blocks of training trials, but this effect was a non-monotonic function of the interval separating trial blocks. Our experiments also revealed that the location of the first block of trials (Room A or Room B) was irrelevant to long-term retention. A memory modulation theoretical framework may provide a useful way to understand these findings. (Contains 8 figures.)