In this thesis, we propose a new distributed run-time environment, which we called cJava, for executing transparently multithread Java applications in clusters. Specifically, cJava is able to hide the physical memory distribution of the cluster's nodes by replacing it with the abstraction of a single distributed shared-memory. To this end, cJava distributes the threads of a Java application over the cluster's nodes in a transparent way to the programmer. Our implementation of cJava required significant extensions to the original Java virtual machine (JVM) to support the abstraction we proposed. First, a distributed object manager was incorporated to the memory management subsystem for creating the global object space. Second, synchronized accesses to the global object space were extended so that they could use the lock() and unlock() primitives the cJava's distributed shared memory system supports. Third, cJava extends the thread subsystem to support remote creation and global monitors. Last, a subsystem for remote signaling was added to the original JVM. The main property of cJava is that it does not require any modification to the Java Language specification. Most importantly, our experimental results of cJava's performance demonstrate that the strategy this thesis defends for building Java clusters is not only feasible but also highly promising.