Abstract: Introduce object-oriented metrics; define the detailed definitions of McCabe, Halstead and weighted methods per class(WMC) for Java; a new method for implementation of measurement tool for Java was offered; demonstration for using this tool to optimize program and enhance software quality was also given. Experiments show that this tool helps programmers to improve program effectively. They also show that the module with high measurement value are inclined to be complex, fallible, hard to maintain and need consume amount of labor, so testers have to pay more attention to test those modules. Developers must re-exam those modules, check their logic, and try to optimize them to depress their measurement value. By providing adequate experiment data, The results show that software measurement plays an essential roll for improving software quality. The metrics used are depth of inheritance tree(DIT), number of children(NOC), McCabe, Halstead, weighted methods per class(WMC) and response for a class(RFC). They help programmers analyze Java program at three aspects——the inheritance metrics, the module complexity metrics, and the communication metrics. So each kind of the metrics measures Java program from different aspects, evaluates the software quality effectively, and reveals the probable bugs.