Race Condition in Concurrent Counter: Lost Increments in Multithreading

I'm using a shared Counter class with counter++ in a multithreaded Java app. Despite starting 100 threads each calling increment() 1000 times, the final count often falls far short of 100,000. Why?

counter++ is not atomic—it's decomposed into read modify write steps. Parallel access without synchronization causes race conditions where increments are lost ([turn0search0]turn0search6]turn0search14]). Fixes: • Mark methods synchronized, or wrap increments inside synchronized blocks. • Use AtomicInteger: java Copy code private final AtomicInteger count = new AtomicInteger(0); count.incrementAndGet(); • Or use thread-safe collections or locks (e.g. ReentrantLock) for managing shared mutable state reliably.