A study was conducted to determine whether cooperative testing would result in better performance and less anxiety than individual testing. Two sections of a developmental psychology class used cooperative testing, and a third section used traditional testing methods. Four multiple choice tests containing 50 items were administered to each group, with items equally divided between easy and moderately difficult items. Each student had his/her own answer sheet which contained items asking for a rating of anxiety and expected performance and an indication of preferred type of testing. Test completion times were noted when students turned in their exam. Several days prior to the first cooperative exam, students were told that they could pick a partner for the next exam, but that they would each turn in their own answer sheets. They were allowed to change partners for the second cooperative examination and were not required to participate. In one class section, 7 of 34 students chose not to participate, while in the second section 9 of 24 chose not to participate. The mean test scores of the non-participating students did not differ significantly from the means of the subjects taking individual tests. At first glance, cooperative testing seemed to produce superior performance; however, not all students benefited equally from cooperative testing. The two cooperative-testing class sections showed very different patterns of performance. Though there were no differences in anxiety between students using cooperative and traditional testing methods, students who used both methods showed a strong preference for cooperative testing. (AC)