Interfirm network expansion and maintenance, as an important aspect of network dynamics, is fundamentally about a firm’s search of new or existing partners. As strategic choices, firms’ partner selections are subject to the influences of endogenous and exogenous factors. From a resource dependence logic, we build a multi-level framework to examine firms’ tendency for choosing new or existing alliance partners, bridging theories from organizational learning, institutional theory, and social networks. Specifically, we focus on the context of strategic alliance networks and examine factors at levels of the firm (internal resource capacity, firm status, and firm structural holes), the environment (munificence and uncertainty), and the network (network stage). We do so with the help of a unique research method that combines empirical investigations of five U.S. industries from 1988 to 1995 with a computer simulation model of dynamic network behaviors. Both empirical and computational findings show that a firm’s internal resource capacity, status, and structural hole positions, as well as the environmental conditions, affect how it expands or maintains its network relations. Further, the simulation model reveals that the impact from firm and environmental level factors can be subject to the constraints of network stage. Our study provides critical evidence into the necessity and viability of adopting a multi-level and multi-method approach in understanding network dynamics.