If there are native methods in the call stack at the time of the exception, there are important restrictions to note about the returned catch location. In such cases, it is not possible to predict whether an exception will be handled by some native method on the call stack. Thus, it is possible that exceptions considered uncaught here will, in fact, be handled by a native method and not cause termination of the target VM. Furthermore, it cannot be assumed that the catch location returned here will ever be reached by the throwing thread. If there is a native frame between the current location and the catch location, the exception might be handled and cleared in that native method instead.

Note that the compiler can generate try-catch blocks in some cases where they are not explicit in the source code; for example, the code generated for synchronized and finally blocks can contain implicit try-catch blocks. If such an implicitly generated try-catch is present on the call stack at the time of the throw, the exception will be considered caught even though it appears to be uncaught from examination of the source code.