ReentrantLock provides the same locking and memory semantics as intrinsic locking, as well as additional features such as timed lock waits, interruptible lock waits, fairness, and the ability to implement non-block-structured locking. The performance of ReentrantLock appears to dominate that of intrinsic locking, winning slightly on Java 6 and dramatically on Java 5.0. So why not deprecate synchronized and encourage all new concurrent code to use ReentrantLock? Some authors have in fact suggested this, treating synchronized as a "legacy" construct. But this is taking a good thing way too far.
Intrinsic locks still have significant advantages over explicit locks. The notation is familiar and compact, and many existing programs already use intrinsic lockingand mixing the two could be confusing and error-prone. Reentrant-Lock is definitely a more dangerous tool than synchronization; if you forget to wrap the unlock call in a finally block, your code will probably appear to run properly, but you've created a time bomb that may well hurt innocent bystanders. Save ReentrantLock for situations in which you need something ReentrantLock provides that intrinsic locking doesn't.
ReentrantLock is an advanced tool for situations where intrinsic locking is not practical. Use it if you need its advanced features: timed, polled, or interruptible lock acquisition, fair queueing, or non-block-structured locking. Otherwise, prefer synchronized.
Under Java 5.0, intrinsic locking has another advantage over ReentrantLock: tHRead dumps show which call frames acquired which locks and can detect and identify deadlocked threads. The JVM knows nothing about which threads hold ReentrantLocks and therefore cannot help in debugging threading problems using ReentrantLock. This disparity is addressed in Java 6 by providing a management and monitoring interface with which locks can register, enabling locking information for ReentrantLocks to appear in thread dumps and through other management and debugging interfaces. The availability of this information for debugging is a substantial, if mostly temporary, advantage for synchronized; locking information in thread dumps has saved many programmers from utter consternation. The non-block-structured nature of ReentrantLock still means that lock acquisitions cannot be tied to specific stack frames, as they can with intrinsic locks.
Future performance improvements are likely to favor synchronized over ReentrantLock. Because synchronized is built into the JVM, it can perform optimizations such as lock elision for thread-confined lock objects and lock coarsening to eliminate synchronization with intrinsic locks (see Section 11.3.2); doing this with library-based locks seems far less likely. Unless you are deploying on Java 5.0 for the foreseeable future and you have a demonstrated need for ReentrantLock's scalability benefits on that platform, it is not a good idea to choose ReentrantLock over synchronized for performance reasons.