An experiment is described which confirms earlier findings that the main difference between recognition and recall in signal-detection terms lies in higher criterion levels for recall. The experiment also illustrates a method of applying signal-detection measures to recall in the very common case, hitherto considered impossible to treat, where the number of potential distractors is not known a priori and false recalls become progressively fewer over a number of trials. The method provides a more sensitive and efficient measure of rote learning than those commonly in use and opens the way to a reappraisal of many traditional assumptions about learning and memory.