Most Hibernate examples are in terms of 2 tables.

However, we already have an existing (read legacy) database. Most of the associations are between atleast 3 tables. As a result the join table/association table consists of 3 FKs from each of the 3 tables, in addition to the association's properties. Moreover, the application is always interested in dealing with the AssocationTable entity and never the primary entities themselves.

Let me give you an example: We have 3 tables as follows:
1. Agency [PK:AgencyID]
2. Agency_Website (An agency can have multiple websites) [PK:SiteID]
3. ThirdPartyVendors (A list of all the 3rd party vendors that the agency can connect to) [PK:VendorID]

Now the 4th table is the Join table - AgencyWebsiteVendorConnection. Its columns are as follows:
1. AgencyID
2. SiteID
3. VendorID
4. ConnectionCredentialKey [can contain values like Username, Userid, accountNo - since different vendors have different names for identifying users]
5. ConnectionCredentialValue [actual values for the above key]

This is the simplest example we have. Most other examples have more than 4 tables and the join exists between 4 or more tables at a time. At that point of time, we tend to loose focus on the multiplicity of the association - we cannot distinguish whether the join table is 1-many, many-many etc.

Can someone please provide an example of how these mappings can be done?

I think the recommendation is to map the association tables as full entities. If it is as you say that the application is mainly interested in managing the associations tables, this will make life a lot easier. The main problem may be that you need composite primary keys, but there are some good examples of that here: http://www.hibernate.org/hib_docs/v3/re ... iteid.html