Although many studies of the processes through which immigrants become integrated into their host societies have emphasized social capital and trust, they have tended to do so only with regard to the immigrant groups themselves; that is, they have examined the effects of social capital and trust within immigrant groups. A theory is offered that posits a positive contextual effect for the aggregate-level social capital and trust existing in the host societies in which the immigrant groups find themselves embedded. This effect is argued to exist independently of the social capital stores of the immigrant groups in question. In order to empirically test this theory (as well as to contribute to rectifying the paucity of empirical studies on integration more generally), a novel measurement of immigrant integration is developed based on the conceptualization of integration as the process by which immigrant populations become similar to the native-born with respect to certain indicators. In order to avoid common statistical assumptions regarding data-generating processes and functional forms, this measurement is generated using a neural network. Alternative hypotheses regarding multicultural policies, welfare state expansiveness, and macroeconomic and geographic determinants are developed and evaluated. A cross-national analysis provides support for the primary hypothesis, but can only be regarded as suggestive due to its data limitations. An analysis across the US states is undertaken to expand the number of available data points and thus enhance the validity of the statistical inferences drawn. In a multilevel analysis across all state-groups, the macroeconomic and geographic controls outperform the other structural variables, as well as social capital. A case study of the Mexico-born in the US, however, eliminates outliers and shows that social capital does indeed have a positive and significant effect even in the presence of the control variables. Implications for current theoretical understanding and policymaking and suggestions for future research are then discussed in light of these findings.