Investigates whether the performance and attitudes of students could be improved by giving them some control over the group selection process. Groups were formed either by randomly combining paired friends or by randomly assigning all students. Students completed a group exercise and a group case. The dependent variables were the project grades and student satisfaction. Student satisfaction was measured using a questionnaire. The results show that attitudes of students were more positive when they were allowed to choose a single friend in the group. The project grades were significantly higher when students were paired, and this result was true regardless of their grade point averages. The interaction between group selection and grade point average is explained, and the paper concludes that the best group selection is to pair friends and then combine them to form groups with high ability rather than randomly assigning students to groups.