From a cluster/SQL setup point of view, adding a third node is not very complicated: you join the new node to the existing cluster and you add a SQL Server instance to an existing cluster. So it's more or less just like adding a second node again.
If you want to create a 3 node cluster it is probably because you want to spare some money in hardware and to have two active nodes that might both failover on the third node. This a correct solution if your SLA states that in case of failover of both active nodes you may have degraded performance.
The challenge here is to define carefully how to share the resources (RAM + CPU) in case you find yourself with, let's say five OLTP instances and two OLAP instances, suddenly running on the same box called Node 3. In the normal situation the 5 OLTP run on Node 1 and you defined how they should share 64GB RAM + 8 cores between them; the two OLAP instances usually run on Node 2 and share (even if it is more difficult to define memory allocation in SSAS) 64GB and 8 cores.
When all that stuff fails over node 3 and there are less than 128GB + 16 cores available on that box, you may have some resource contention problems, right?
In addition to this you might want to prepare all possible scenarios, like all nodes failing on node 1, etc.
The more active nodes you add, the more complexity you get.
MCITP Database administrator 2008
MCTS SQL Server 2008 Implementation and maintenance
MCTS Sharepoint configuration
MCP Designing Deploying and Managing a Network Solution for the Small and Medium-sized Business
ITIL V3 Foundation