Taking a test has been shown to produce enhanced retention of the retrieved information. On tests, however, students often encounter questions the answers for which they are unsure. Should they guess anyway, even if they are likely to answer incorrectly? Or are errors engrained, impairing subsequent learning of the correct answer? We sought to answer this question in 3 experiments. In Experiments 1 and 2, subjects read 80 obscure facts (e.g., "Where is Disko Island? Greenland") and then took a cued recall test. When a subject reported being unable to answer a question, on a randomly chosen half of those questions the computer program insisted upon a guess. Corrective feedback was provided either immediately (Experiment 1) or after a delay (Experiment 2). Forced guessing did not affect subjects' performance on a final test given 1 day later. We extended the investigation to more complex material in Experiment 3. Subjects saw a question (e.g., "Why do ice cubes often pop as they melt in your drink?") and its answer, but for half of the questions, subjects did not see the answer until they first provided a plausible explanation. On a test administered either on the same day or 1 week later, recall performance was again unaffected by a prior wrong guess. (Contains 3 footnotes, 5 tables,and 1 figure.)