We combine economic and institutional theories of clustering in foreign entry location choice in an overarching social learning conceptualization. Prospective entrants learn about the attractiveness of alternative locations by observing the entry choices of previous investors ('models'). We distinguish two types of learning that differ in observational focus width but can and do operate simultaneously. With assessment learning, firms judge the economic feasibility and agglomeration benefits of entering a location by observing and following a broad set of models. With bandwagon learning, firm-level uncertainty narrows attention to, and prompts the following of, specific models, with recentness of model behavior an important moderator. We find broad support for our conceptualization in an analysis of the entries of 692 Japanese electronics firms into Chinese provinces during 1979-2001.