The memory loss that follows a series of electroconvulsive therapy (ECT) treatments has been well documented. The amnesia appears to involve an impairment in the ability to acquire new memories, an impairment of memory for events that occurred shortly before ECT (Dornbush, 1972; Williams, 1966), and an impairment in the ability to recall material from remote memory (Janis, 1950; Squire, 1974a). Several characteristics of the anterograde amnesia produced by ECT have been the subject of contradictory reports. For example, it has been reported that memory functions can improve markedly or even return to pre-ECT levels within hours after the last of a short series of treatments (Brengelmann, 1959; Zinkin and Birtchnell, 1968; Zirkle, 1956). Yet, it has also been found that amnesic effects of a short series of treatments can be detected for weeks (Cronin, Bodley, Mather, Gardner and Tobin, 1970; Halliday, Davison, Browne and Krieger, 1968). Moreover, some reports suggest that memory recovers at the same rate following each of the first few treatments in a series (Brengelmann, 1959; Zinkin and Birtchnell, 1968), whereas other reports demonstrate that the effects of the first few treatments on memory can be cumulative (Bidder, Strain and Brunschwig, 1970). One explanation for such variant findings is that different memory tests have been employed in these studies. To obtain accurate estimates of the effects of ECT on memory, tests used to assess memory should be as sensitive as they can be to memory impairment. This point is particularly relevant, of course, to the evaluation of long-term effects of ECT on memory. Patients who had sustained partial temporal lobectomy could perform normally in tests of immediate reproduction from memory, but were markedly impaired in tests that imposed long delays between learning and retention (Milner, 1958). Apparently tests involving delayed reproduction from memory are a more sensitive index of organic memory dysfunction than tests that involve immediate reproduction. The present study assessed the ability of patients receiving ECT to retain newly learned material for 30 minutes and 24 hours. The results of these delayed retention tests demonstrate that recovery from anterograde amnesia is quite gradual and that the effects of the first few ECT treatments on memory are cumulative; and underscore the value of such tests as sensitive indicators of possible long-term effects of ECT on memory.