Java multithreading allows programs to run multiple tasks concurrently, boosting performance on multi-core processors. Threads share memory space, enabling parallel execution for CPU-intensive work while keeping applications responsive. The core distinction lies between concurrency (interleaved execution on one core) and true parallelism (simultaneous execution across multiple cores).

Developers create threads primarily by implementing the Runnable interface or extending the Thread class. The preferred method uses Runnable, promoting better design by separating task definition from execution. Once started, threads enter defined states like NEW, RUNNABLE, BLOCKED, WAITING, TIMED_WAITING, or TERMINATED, managed by the JVM's thread scheduler.

Proper synchronization is critical when threads access shared resources. Without it, race conditions and memory consistency errors can corrupt data. Java provides intrinsic locks (monitors) on objects. The `synchronized` keyword, used on methods or blocks, ensures only one thread executes critical sections at a time, maintaining thread safety.