Abstract
The software reliability plays an important role nowadays. From year to year the systems become more and more complex. Consequently, the software becomes more and more complex, too and this can cause more software errors. In this paper a new method is introduced, which serves to predict the reliability of software in a better way. Furthermore, the presumption that the hazard rate is constant is rebutted and a more realistic assumption is made. It is demonstrated that by the use of the new approach the prediction of the number of remaining errors is possible in a better way.