In this analysis we present a whole population network which uses administrative data to construct a network incorporating 1.4 billion relationships between the 17 million inhabitants of the Netherlands. Relationships are identified between individuals who live in the same household, live close to each other, work for the same company, attend the same educational institution, or belong to the same extended family. This network has properties that are rare in observed social networks, which opens up new applications for network science in the social sciences. To demonstrate the applications of such a network, we use a random walk approach to estimate exposure to individuals of differing educational backgrounds and whether specific types of relationships increase or decrease this segregation. The results suggest that household relationships greatly increase segregation whilst work, school and neighborhood networks relationships increase exposure to individuals with different backgrounds. The size of these effects is context dependent. Further applications of a whole population network are also discussed.