In a natural field experiment with 900+ subjects in 300+ teams, we study the effects of choosing team members versus ideas on entrepreneurial team performance. We use a two-by-two design in which subjects are randomly assigned to one of four treatments in which they (i) choose their own team but not the idea they pursue; (ii) choose their own idea but not their team; (iii) choose both their team and the idea to pursue; or (iv) choose neither their team nor the idea. We find that teams who choose their own idea but not their members perform consistently better than those in the remaining treatments. We then explore a number of different channels that can account for this finding.