Abstract: Computational grid is a more variable and unreliable computing environment. It should provide appropriate availability assurance for the applications. How to measure the availability is the key issue for providing high availability service. In this paper, the characteristics of the availability of computational grid are analyzed and it is proposed that the availability of computational grid should be measured from the applications’ point of view. Then a measurement model for the availability of applications in the computational grid is presented. In this model the overall availability of a grid application is measured by using a serial RBD, and the availability of application’s tasks on a grid node is measured by using a probabilistic model. This measurement model abstracts the key factors which affect the availability of applications in the computational grid. Then the usage of this measurement model in the high availability service architecture of the computational grid is discussed.