Microfinance institutions now serve over 10 million poor households in the developing and developed world, and much of their success has been attributed to their innovative use of peer group lending. There is very little empirical evidence, however, to suggest that group lending schemes offer a superior institutional design over lending programs that serve individual borrowers. The authors find empirical evidence that group lending does indeed lower borrower default rates more than conventional individual lending, and that this effect operates through the dual channels of selection into the peer lending program and, once inside the program, greater group borrower effort.