Grid computing is a type of distributed computing that has shown promising applications in many fields. A great concern in grid computing is the cheating problem described in the following: a participant is given D = {x1, . . . , xn}, it needs to compute f(x) for all x ∊ D and return the results of interest to the supervisor. How does the supervisor efficiently ensure that the participant has computed f(x) for all the inputs in D, rather than a subset of it? If participants get paid for conducting the task, there are incentives for cheating. In this paper, we propose a novel scheme to achieve the uncheatable grid computing. Our scheme uses a sampling technique and the Merkle-tree based commitment technique to achieve efficient and viable uncheatable grid computing.