Basically you define all your hosts/services on a "Central Server". The Central Server need not do any monitoring, but it can perform checks an a subset of nodes. The Central Server is configured for passiv checks for a set of nodes and services monitored by another instance of Nagios. This "Distributed Server" submits all it's results to the "Central Server". This gives you a single cohesive view of your network.
It's up to you whether you fell there is an advantage to doing this. Personally I'd rather have several instances of Nagios and check each independently of the others.