We propose a probabilistic model of integration where decisions about integration are made based on the perceived ensuing benefit. Integration proceeds in a pairwise manner. This model provides a way to think formally about the integration processes in networks. The model considers information decay over time, and relates decisions to integrate to the value of the information the nodes possess. A sensor network designed to detect intruders is used as an illustrative example. Other applications of the model are suggested: information systems integration, as well as the merger of corporations.