This paper addresses task-allocation problems with uncertainty in situational awareness for distributed autonomous robots (DARs). The uncertainty propagation over a task-allocation process is done by using the Unscented transform that uses the Sigma-Point sampling mechanism. It has great potential to be employed for generic task-allocation schemes, in the sense that there is no need to modify an existing task-allocation method that has been developed without considering the uncertainty in the situational awareness. The proposed framework was tested in a simulated environment where the decision-maker needs to determine an optimal allocation of multiple locations assigned to multiple mobile flying robots whose locations come as random variables of known mean and covariance. The simulation result shows that the proposed stochastic task allocation approach generates an assignment with 30% less overall cost than the one without considering the uncertainty.