Imagine administering an online standardized test to an entire class of 11th-grade students when, halfway through the exam, the server holding the test hits a snag and throws everyone offline. Imagine another scenario in which an elementary school has very few computers so teachers must bus their students to the local high school for a timed test. At the new test site, six students suddenly refuse to take the test and begin crying, while simultaneously more than half of the students discover they have trouble reaching the keyboards. These scenarios demonstrate that there are many risks in using computers to test students. However, thinking ahead about the needs of students, the testing site and the system being implemented limits the potential for problems to arise during testing. In this article, these risks are discussed and possible solutions are suggested.