This paper reports the results of a laboratory experiment on network formation among heterogeneous agents. The experimental design extends the basic Bala-Goyal (2000) model of network formation with decay and two-way flow of benefits by allowing for agents with lower linking costs or higher benefits to others. We consider treatments where agents' types are common knowledge and treatments where agents’ types are private information. In all treatments, the (efficient) equilibrium network has a "star" structure. We find that with homogeneous agents, equilibrium predictions fail completely. In contrast, with heterogeneous agents stars frequently occur, often with the high-value or low-cost agent in the center. Stars are not borne but rather develop: in treatments with a high-value agent, the network's centrality, stability, and efficiency all increase over time. Our results suggest that agents' heterogeneity is a major determinant for the predominance of star-like structures in real-life social networks.