We use recruitment into a laboratory experiment in Kolkata, India to analyze how social networks select individuals for jobs. The experiment allows subjects to refer actual network members for casual jobs as experimental subjects under exogenously varied incentive contracts. We provide evidence that some workers, those who are high ability, have useful information about the abilities of members of their social network. However, the experiment also shows that social networks provide incentives to refer less qualified workers, and firms must counterbalance these incentives in order to effectively use existing employees to help overcome their screening problem.