This paper uses group attachment theory and social defense theory to understand how workgroup attachment security affects cross-group collaboration performance. 675 MBAs in the United States participated in a simulation of cross-group collaboration in which their task was to manufacture a new telecommunication device. Participants were randomly assigned to either a design or a manufacturing team and those participants in each team were further randomly assigned to either group-attachment-security prime or control condition. Results indicate that cross-functional task accuracy was higher when the design team was primed with a group attachment security (vs. in the control condition). And, cross-functional task speed was slower when both the design team and the manufacturing team were primed with a group attachment security than either or neither team was primed with a group attachment security. Thus, this study highlights that there are benefits (task accuracy) as well as costs (task speed) in securely attached workgroups.