A novel stochastic version of the orienteering problem is considered. The problem generalizes the probabilistic TSP. A deterministic equivalent MILP model is derived. A branch-and-cut procedure with special branching rules is developed. Matheuristic procedures are proved to be efficient and effective.