The potential drawbacks of having multiple datafiles are trivial compared to the potential benefits.

On most platforms when you issue a CREATE TABLESPACE command with multiple datafiles, those datafiles will be "zeroed" in parallel by slave threads. Even on platforms that do not support this, you can do multiple ALTER TABLESPACE ADD DATAFILE commands in parallel. (This ability was added in version 7.1.4 in response to bug 210833.) Therefore multiple datafiles can make tablespace creation much faster.

For file system based databases, having multiple datafiles may reduce the risk of contention for the operating system's datafile read/write locks (inode locks).

Having multiple datafiles per tablespace facilitates the adoption of a uniform datafile size policy, which in turn facilitates disk load rebalancing.

Finally, I answered a related question on Ixora Answers a few weeks ago, and you may find that answer relevant. Look for the answer headed Is AUTOEXTEND OK?.

Could any one point out the advantages/disadvantages of having multiple datafiles for a tablespace as against having one datafile per tablespace?