Computational grids comprise heterogeneous collections of resources. Resources have varying and unreliable performance because of network bandwidth, system load, network failure, connection between grid resources, tasks running on resources and various other factors. To predict when a new resource will be added or removed is quite difficult. A resource that was available at certain time might not be available in future. The resources administered under different policies in the grid forces resources having different availability properties. It also makes predicting the availability and selection of optimized resources a challenging job. In this paper, an attempt is made to survey few methods for prediction of resource availability in grids. © 2014 IEEE.