Two experiments were conducted to determine whether differences in expectations of tests on specific information items or on interrelationships among those items would lead to differences in performances on the two types of tests. In one experiment 30 undergraduates were led by instructions and prior experience to expect either free recall or serial recall of word lists. Test expectancy effects were obtained for both types of tests (p<.01). In the second experiment no test expectancy results were obtained for 60 undergraduates expecting to be tested on cither specific statements or logical implications of those statements from textual materials. The contrasting results may imply that test expectancy effects occur only when tests represent clearly conflicting task requirements, though methodological difference between experiments provides an alternative explanation.