This paper addresses two important questions concerning social fragmentation in work teams. First, from where do disconnections between team members, measured in terms of the proportion of structural holes within the work team, derive? Second, what are the consequences for team performance of having more or less structural holes between team members? In answering the first question, the research investigated whether demographic diversity in teams played a role in predicting the proportion of structural holes in team friendship networks. For 19 teams at a wood products company, there were no effects of ethnic and gender diversity on structural hole proportions. However, age diversity significantly reduced the extent of structural ‘holeyness.’ In investigating the second question, two countervailing tendencies were considered. In the absence of structural holes, teams are likely to be at low risk for new ideas. But fragmented teams in which team members are separated by many structural holes are likely to have difficulty coordinating. The researchers demonstrated a curvilinear effect: a moderate level of structural diversity in teams was positively associated with team performance. Thus, the research suggested that it is structural diversity (measured in terms of the proportion of structural holes) rather than demographic diversity that matters in the prediction of team performance. Copyright © 2006 John Wiley & Sons, Ltd.