Following practice lists for recall and recognition, a third list was given for which half the third- and sixth-grade children expected a recall test and half a recognition test. All children then were given a recall test followed by a recognition test. Recall by sixth graders was better when a recall rather than a recognition test was expected. Test expectancy did not affect the recall of third graders, nor did it apparently affect recognition at either grade level. Self-reports indicate greater preexperimental understanding of task difficulty differences at Grade 6. At Grade 6, but not at Grade 3, reports of rehearsal of items in blocks (rather than single-item repetition) were more frequent when a recall test was expected and reports of image production were more frequent when a recognition test was expected.