The elective surgery planning problem for operating rooms shared between elective and emergency patients is addressed. The planning problem consists in determining the set of elective patients to be operated on in each operating room in each period over a planning horizon in order to minimize patient-related costs and the expected operating rooms' utilization costs. A stochastic mathematical programming model and a column generation approach are proposed. The proposed approach results in both a near-optimal solution and a lower bound to assess the degree of optimality. Solutions within 2% of the optimum are obtained in a short computation time for problems of practical sizes with 12 operating rooms and about 210 elective patients.